diff --git a/.env.example b/.env.example
new file mode 100644
index 0000000..ace6c1b
--- /dev/null
+++ b/.env.example
@@ -0,0 +1,7 @@
+FLASK_APP=app
+FLASK_ENV=development
+OLLAMA_HOST=http://localhost:11434
+LLM_MODEL=llama3.2:latest
+LLM_CONTEXT=4096
+LLM_MAX_TOKENS=768
+LLM_TEMPERATURE=0.2
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..6c1f29d
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,12 @@
+.venv/
+__pycache__/
+*.pyc
+instance/
+.env
+uploads/*
+!uploads/.gitkeep
+*.log
+.DS_Store
+node_modules/
+client/node_modules/
+server/node_modules/
diff --git a/README.md b/README.md
index 3a08a72..b588e37 100644
--- a/README.md
+++ b/README.md
@@ -1 +1,128 @@
-# beta
\ No newline at end of file
+# Aurora CRM Intelligence Suite
+
+Aurora CRM Intelligence Suite is a local-first command center for predictive strategy, risk analytics, AI advisory chat, and deep document comparison. Every insight runs on your workstation through the [Ollama](https://ollama.com) runtime, so nothing ever leaves your RTX 4090 (or compatible) GPU.
+
+> **Security reminder**
+> No cloud API keys are required. Keep your model checkpoints private and stored locally.
+
+## Project structure
+
+```
+.
+├── app/                  # Flask application package
+│   ├── services/         # Local inference + file helper utilities
+│   ├── static/           # Custom CSS and interaction scripts
+│   └── templates/        # Multi-page Jinja UI
+├── uploads/              # Local document catalog (created automatically)
+├── app.py                # Flask entry point
+├── requirements.txt      # Python dependencies (Flask + requests)
+├── install.ps1           # Windows installer shim (PowerShell)
+├── run.ps1               # Windows auto runner
+├── install.bat           # Convenience launcher that calls install.ps1
+├── run.bat               # Convenience launcher that calls run.ps1
+└── README.md             # You are here
+```
+
+## Prerequisites
+
+- Python **3.10+** (tested with CPython 3.12)
+- PowerShell 5.1+ (bundled with Windows 11) for the automation scripts
+- NVIDIA GPU with recent drivers (4090 recommended) or enough CPU cores if you prefer CPU inference
+- [Ollama for Windows](https://ollama.com/download/windows) installed and running (`ollama serve` starts automatically once installed)
+- A local **GGUF** checkpoint available to Ollama (use `ollama pull llama3.2` or `ollama pull qwen2.5-coder` etc.)
+
+## Quick start (Windows 11)
+
+1. Install [Ollama](https://ollama.com/download/windows) if you have not already, then open **PowerShell** and pull a chat-capable model (for example Meta's Llama 3.2):
+   ```powershell
+   ollama pull llama3.2
+   ```
+   Ollama automatically stores the GGUF checkpoint in its managed directory and serves it to local clients.
+2. Open **PowerShell** and `cd` into the project directory.
+3. If this is your first time running local scripts, allow the session to run them:
+   ```powershell
+   Set-ExecutionPolicy -Scope Process RemoteSigned
+   ```
+4. Run the installer:
+   ```powershell
+   ./install.ps1
+   ```
+   The script will:
+   - Detect Python 3
+   - Create (or reuse) a `.venv` virtual environment
+   - Install dependencies from `requirements.txt`
+5. Create your `.env` file and set the Ollama options:
+   ```powershell
+   Copy-Item .env.example .env
+   notepad .env
+   ```
+   Update `LLM_MODEL` with the model name you pulled (for example `llama3.2:latest`). Leave `OLLAMA_HOST` at the default unless you run Ollama on a different machine.
+6. Launch the stack:
+   ```powershell
+   ./run.ps1
+   ```
+   The runner activates the virtual environment, exports variables from `.env`, verifies the model path, and starts Flask in debug mode at <http://127.0.0.1:5000/>.
+
+You can double-click `install.bat` or `run.bat` from Explorer—they simply relay to the PowerShell scripts with execution policy bypassed.
+
+## Quick start (macOS/Linux)
+
+```bash
+python3 -m venv .venv
+source .venv/bin/activate
+pip install -r requirements.txt
+cp .env.example .env  # edit with your Ollama host and model
+export OLLAMA_HOST="http://localhost:11434"
+export LLM_MODEL="llama3.2:latest"
+flask --app app --debug run
+```
+
+Visit <http://127.0.0.1:5000/> once the server reports it is running.
+
+## Feature tour
+
+- **Overview dashboard** – animated chrome with catalog health, workflow readiness, and immersive storytelling.
+- **Data steward** – upload TXT, MD, CSV, or JSON files and request local AI summaries with quality scoring, tags, and opportunity prompts.
+- **Predictive strategy** – craft revenue scenarios, score growth outlooks, and review initiative guidance with progress dials and impact distributions.
+- **Risk command** – quantify exposure, inspect mitigation cards, and monitor early warnings with visual coverage bars.
+- **Advisor lounge** – chat with the embedded strategist, auto-load document summaries, and keep every conversation on device.
+- **Comparison lab** – run deep comparisons between two documents to expose alignment, divergence, and actionable recommendations.
+
+Each workflow uses structured prompts defined in `app/services/insights.py`. Responses are parsed as JSON and rendered dynamically with a custom glassmorphism UI—no external CDNs required.
+
+## Configuration notes
+
+- `OLLAMA_HOST` – Base URL where Ollama listens (default `http://localhost:11434`).
+- `LLM_MODEL` – **required**. Model name as listed by `ollama list` (for example `llama3.2:latest`).
+- `LLM_CONTEXT` – Context window supplied to Ollama (default `4096`). Increase if your model supports a larger context.
+- `LLM_MAX_TOKENS` – Maximum tokens to generate per response (default `768`).
+- `LLM_TEMPERATURE` – Sampling temperature (default `0.2`).
+- Uploaded documents remain on disk inside the local `uploads/` folder. Delete files there to clear the catalog.
+
+## Automation scripts
+
+The Windows automation lives under `scripts/windows/` and is intentionally PowerShell-only—no external tooling required.
+
+- `install.ps1` – ensures Python is present, provisions `.venv`, and installs dependencies.
+- `run.ps1` – verifies the environment, loads `.env`, reminds you if Ollama variables are unset, and starts Flask using the virtual environment's interpreter.
+
+Both scripts print friendly diagnostics (missing Python, missing model, etc.) so issues are easy to resolve.
+
+## Validation checklist
+
+- Flask routes use built-in templating; no frontend framework is required.
+- All inference stays local by calling the Ollama HTTP API on `OLLAMA_HOST`; there are **no** outbound internet calls from the Flask app.
+- File uploads are limited to text-based formats and capped at 10 MB per file to keep processing light.
+- UI assets rely entirely on bundled CSS/JS with layered gradients, animated nav, and custom progress visuals.
+
+## Troubleshooting
+
+| Symptom | Resolution |
+| --- | --- |
+| "Local inference is not configured" error | Launch Ollama (it starts automatically after installation), run `ollama pull <model>`, and set `LLM_MODEL` in `.env`. |
+| App loads but responses are empty | Confirm the pulled Ollama model supports chat instruction tuning and that `LLM_CONTEXT` is within the model's limit. |
+| Slow responses on Windows | Verify Ollama is using your GPU (run `ollama run llama3.2 "test"` in a console and watch GPU usage). |
+| PowerShell blocks the scripts | Run PowerShell as Administrator once and execute `Set-ExecutionPolicy -Scope Process RemoteSigned`. |
+| Installer reports Python not found | Install Python 3.10+ from <https://www.python.org/downloads/> and tick "Add to PATH" during setup. |
+
+Enjoy orchestrating predictive CRM intelligence locally!
diff --git a/app.py b/app.py
new file mode 100644
index 0000000..488dae9
--- /dev/null
+++ b/app.py
@@ -0,0 +1,6 @@
+from app import create_app
+
+app = create_app()
+
+if __name__ == "__main__":
+    app.run(debug=True)
diff --git a/app/__init__.py b/app/__init__.py
new file mode 100644
index 0000000..7ea159b
--- /dev/null
+++ b/app/__init__.py
@@ -0,0 +1,18 @@
+from flask import Flask
+from pathlib import Path
+
+from dotenv import load_dotenv
+
+
+def create_app() -> Flask:
+    load_dotenv()
+    app = Flask(__name__)
+    app.config["UPLOAD_FOLDER"] = Path(app.root_path).parent / "uploads"
+    app.config["UPLOAD_FOLDER"].mkdir(parents=True, exist_ok=True)
+    app.config["MAX_CONTENT_LENGTH"] = 10 * 1024 * 1024  # 10 MB per file
+    app.config["SECRET_KEY"] = app.config.get("SECRET_KEY") or "dev-secret-key"
+
+    from . import routes  # noqa: WPS433
+
+    routes.register(app)
+    return app
diff --git a/app/routes.py b/app/routes.py
new file mode 100644
index 0000000..9b78343
--- /dev/null
+++ b/app/routes.py
@@ -0,0 +1,234 @@
+from __future__ import annotations
+
+import json
+from collections import Counter
+from typing import Any, Dict, List, Tuple
+
+from flask import (
+    Blueprint,
+    Response,
+    current_app,
+    flash,
+    jsonify,
+    redirect,
+    render_template,
+    request,
+    url_for,
+)
+from werkzeug.utils import secure_filename
+
+from .services import files as file_service
+from .services.insights import InsightService, InsightServiceError
+
+blueprint = Blueprint("core", __name__)
+
+insight_service = InsightService()
+
+
+@blueprint.app_errorhandler(InsightServiceError)
+def handle_insight_error(error: InsightServiceError) -> tuple[Response, int]:
+    return (
+        jsonify({
+            "error": str(error),
+            "hint": "Start Ollama, pull your model, and ensure OLLAMA_HOST/LLM_MODEL are configured.",
+        }),
+        400,
+    )
+
+
+@blueprint.route("/")
+def overview() -> str:
+    return _render_workspace("overview.html")
+
+
+@blueprint.route("/data-management")
+def data_management() -> str:
+    return _render_workspace("data_management.html")
+
+
+@blueprint.route("/predictive-strategy")
+def predictive_strategy() -> str:
+    return _render_workspace("prediction.html")
+
+
+@blueprint.route("/risk-command")
+def risk_command() -> str:
+    return _render_workspace("risk.html")
+
+
+@blueprint.route("/ai-advisor")
+def ai_advisor() -> str:
+    return _render_workspace("advisor.html")
+
+
+@blueprint.route("/comparison-lab")
+def comparison_lab() -> str:
+    return _render_workspace("comparison.html")
+
+
+@blueprint.route("/data/upload", methods=["POST"])
+def upload_file() -> Response:
+    if "document" not in request.files:
+        flash("Select a document before uploading.", "danger")
+        return redirect(url_for("core.data_management"))
+
+    document = request.files["document"]
+    if not document.filename:
+        flash("The selected document has no filename.", "danger")
+        return redirect(url_for("core.data_management"))
+
+    filename = secure_filename(document.filename)
+    try:
+        saved_path = file_service.save_upload(
+            storage=document,
+            filename=filename,
+            destination=current_app.config["UPLOAD_FOLDER"],
+        )
+    except ValueError as exc:
+        flash(str(exc), "danger")
+        return redirect(url_for("core.data_management"))
+
+    flash(f"Uploaded {filename} to {saved_path.name}.", "success")
+    return redirect(url_for("core.data_management"))
+
+
+@blueprint.route("/api/catalog")
+def api_catalog() -> Response:
+    catalog = file_service.catalog(current_app.config["UPLOAD_FOLDER"])
+    return jsonify({"documents": catalog})
+
+
+@blueprint.route("/api/data/summarize", methods=["POST"])
+def api_summarize() -> Response:
+    payload: Dict[str, Any] = request.get_json(force=True)
+    filename = payload.get("filename")
+    if not filename:
+        return jsonify({"error": "filename is required"}), 400
+
+    try:
+        document = file_service.load_document(current_app.config["UPLOAD_FOLDER"], filename)
+    except (FileNotFoundError, ValueError) as exc:
+        return jsonify({"error": str(exc)}), 400
+
+    summary = insight_service.summarize_document(document)
+    return jsonify(summary)
+
+
+@blueprint.route("/api/predict", methods=["POST"])
+def api_predict() -> Response:
+    data = _clean_payload(request.get_json(force=True))
+    result = insight_service.generate_prediction(data)
+    return jsonify(result)
+
+
+@blueprint.route("/api/risk", methods=["POST"])
+def api_risk() -> Response:
+    data = _clean_payload(request.get_json(force=True))
+    result = insight_service.assess_risk(data)
+    return jsonify(result)
+
+
+@blueprint.route("/api/advice", methods=["POST"])
+def api_advice() -> Response:
+    data = _clean_payload(request.get_json(force=True))
+    result = insight_service.provide_advice(data)
+    return jsonify(result)
+
+
+@blueprint.route("/api/chat", methods=["POST"])
+def api_chat() -> Response:
+    data = _clean_payload(request.get_json(force=True))
+    conversation: List[Dict[str, str]] = data.get("conversation", [])
+    if not isinstance(conversation, list):
+        return jsonify({"error": "conversation must be a list"}), 400
+
+    result = insight_service.chat(conversation, context=data.get("context"))
+    return jsonify(result)
+
+
+@blueprint.route("/api/compare", methods=["POST"])
+def api_compare() -> Response:
+    data = _clean_payload(request.get_json(force=True))
+    left_name = data.get("left")
+    right_name = data.get("right")
+    if not left_name or not right_name:
+        return jsonify({"error": "left and right filenames are required"}), 400
+
+    try:
+        left_doc = file_service.load_document(current_app.config["UPLOAD_FOLDER"], left_name)
+        right_doc = file_service.load_document(current_app.config["UPLOAD_FOLDER"], right_name)
+    except (FileNotFoundError, ValueError) as exc:
+        return jsonify({"error": str(exc)}), 400
+    result = insight_service.compare_documents(left_doc, right_doc)
+    return jsonify(result)
+
+
+def _clean_payload(payload: Dict[str, Any] | None) -> Dict[str, Any]:
+    if payload is None:
+        return {}
+    return json.loads(json.dumps(payload))
+
+
+def register(app) -> None:  # pragma: no cover - flask wiring
+    app.register_blueprint(blueprint)
+
+
+def _render_workspace(template: str, **context: Any) -> str:
+    catalog, catalog_summary, recent_documents = _build_catalog_context()
+    return render_template(
+        template,
+        catalog=catalog,
+        catalog_summary=catalog_summary,
+        recent_documents=recent_documents,
+        api_ready=insight_service.available,
+        **context,
+    )
+
+
+def _build_catalog_context() -> Tuple[List[Dict[str, Any]], Dict[str, Any], List[Dict[str, Any]]]:
+    catalog = file_service.catalog(current_app.config["UPLOAD_FOLDER"])
+    summary = _summarize_catalog(catalog)
+    spotlight = catalog[:3]
+    return catalog, summary, spotlight
+
+
+def _summarize_catalog(catalog: List[Dict[str, Any]]) -> Dict[str, Any]:
+    if not catalog:
+        return {
+            "count": 0,
+            "total_size": "0 B",
+            "average_size": "0 B",
+            "last_updated": "—",
+            "coverage": [],
+        }
+
+    total_bytes = sum(item.get("size_bytes", 0) for item in catalog)
+    average_bytes = total_bytes / len(catalog)
+    latest = max(catalog, key=lambda item: item.get("updated_epoch", 0))
+    coverage_counter = Counter(item.get("type", "OTHER") for item in catalog)
+    coverage = [
+        {
+            "label": label,
+            "value": count,
+            "percent": round((count / len(catalog)) * 100),
+        }
+        for label, count in coverage_counter.most_common()
+    ]
+
+    return {
+        "count": len(catalog),
+        "total_size": _format_bytes(total_bytes),
+        "average_size": _format_bytes(int(average_bytes)),
+        "last_updated": latest.get("updated"),
+        "coverage": coverage,
+    }
+
+
+def _format_bytes(value: int) -> str:
+    if value < 1024:
+        return f"{value} B"
+    if value < 1024 ** 2:
+        return f"{value / 1024:.1f} KB"
+    if value < 1024 ** 3:
+        return f"{value / (1024 ** 2):.1f} MB"
+    return f"{value / (1024 ** 3):.1f} GB"
diff --git a/app/services/files.py b/app/services/files.py
new file mode 100644
index 0000000..09ad586
--- /dev/null
+++ b/app/services/files.py
@@ -0,0 +1,96 @@
+from __future__ import annotations
+
+import csv
+import io
+import json
+from datetime import datetime
+from pathlib import Path
+from typing import BinaryIO, Dict, List
+
+ALLOWED_EXTENSIONS = {"txt", "md", "csv", "json"}
+
+
+class Document(Dict[str, str]):
+    """Typed dictionary representing an uploaded document."""
+
+
+def catalog(directory: Path) -> List[Dict[str, str]]:
+    directory.mkdir(parents=True, exist_ok=True)
+    entries: List[Dict[str, str]] = []
+    for file_path in sorted(
+        directory.glob("*"), key=lambda path: path.stat().st_mtime, reverse=True
+    ):
+        if not file_path.is_file():
+            continue
+        size_bytes = file_path.stat().st_size
+        extension = file_path.suffix.lstrip(".").upper() or "N/A"
+        updated_dt = datetime.fromtimestamp(file_path.stat().st_mtime)
+        entries.append(
+            {
+                "name": file_path.name,
+                "size": _format_size(size_bytes),
+                "size_bytes": size_bytes,
+                "type": extension,
+                "updated": updated_dt.strftime("%Y-%m-%d %H:%M"),
+                "updated_iso": updated_dt.isoformat(),
+                "updated_epoch": file_path.stat().st_mtime,
+            }
+        )
+    return entries
+
+
+def save_upload(storage: BinaryIO, filename: str, destination: Path) -> Path:
+    extension = filename.rsplit(".", 1)[-1].lower()
+    if extension not in ALLOWED_EXTENSIONS:
+        raise ValueError(
+            f"Unsupported file type: .{extension}. Allowed types are {', '.join(sorted(ALLOWED_EXTENSIONS))}."
+        )
+    destination.mkdir(parents=True, exist_ok=True)
+    file_path = destination / filename
+    if hasattr(storage, "seek"):
+        storage.seek(0)
+        data = storage.read()
+    else:
+        storage.stream.seek(0)
+        data = storage.stream.read()
+    with open(file_path, "wb") as buffer:
+        buffer.write(data)
+    return file_path
+
+
+def load_document(directory: Path, filename: str) -> Document:
+    file_path = (directory / filename).resolve()
+    if not file_path.exists():
+        raise FileNotFoundError(f"Document {filename} was not found in uploads directory.")
+    text = _extract_text(file_path)
+    return Document(
+        name=filename,
+        text=text,
+        size=str(file_path.stat().st_size),
+        type=file_path.suffix.lstrip(".") or "unknown",
+    )
+
+
+def _extract_text(file_path: Path) -> str:
+    extension = file_path.suffix.lower()
+    if extension in {".txt", ".md"}:
+        return file_path.read_text(encoding="utf-8", errors="ignore")
+    if extension == ".json":
+        data = json.loads(file_path.read_text(encoding="utf-8", errors="ignore"))
+        return json.dumps(data, indent=2)
+    if extension == ".csv":
+        content = io.StringIO(file_path.read_text(encoding="utf-8", errors="ignore"))
+        reader = csv.reader(content)
+        rows = [", ".join(row) for row in reader]
+        return "\n".join(rows)
+    raise ValueError(f"Unable to extract text from {file_path.name}.")
+
+
+def _format_size(size_bytes: int) -> str:
+    if size_bytes < 1024:
+        return f"{size_bytes} B"
+    if size_bytes < 1024 ** 2:
+        return f"{size_bytes / 1024:.1f} KB"
+    if size_bytes < 1024 ** 3:
+        return f"{size_bytes / (1024 ** 2):.1f} MB"
+    return f"{size_bytes / (1024 ** 3):.1f} GB"
diff --git a/app/services/insights.py b/app/services/insights.py
new file mode 100644
index 0000000..e016877
--- /dev/null
+++ b/app/services/insights.py
@@ -0,0 +1,213 @@
+from __future__ import annotations
+
+import json
+import os
+import re
+from typing import Any, Dict, Iterable, List, Optional
+
+import requests
+
+DEFAULT_HOST = os.getenv("OLLAMA_HOST", "http://localhost:11434").rstrip("/")
+DEFAULT_MODEL = os.getenv("LLM_MODEL", "").strip()
+DEFAULT_CONTEXT = int(os.getenv("LLM_CONTEXT", "4096"))
+DEFAULT_MAX_TOKENS = int(os.getenv("LLM_MAX_TOKENS", "768"))
+DEFAULT_TEMPERATURE = float(os.getenv("LLM_TEMPERATURE", "0.2"))
+
+
+class InsightServiceError(RuntimeError):
+    """Raised when the AI assistant cannot complete a request."""
+
+
+class InsightService:
+    def __init__(
+        self,
+        *,
+        host: str = DEFAULT_HOST,
+        model: str = DEFAULT_MODEL,
+        context: int = DEFAULT_CONTEXT,
+        max_tokens: int = DEFAULT_MAX_TOKENS,
+        temperature: float = DEFAULT_TEMPERATURE,
+    ) -> None:
+        self.host = host or "http://localhost:11434"
+        self.model = model
+        self.context = context
+        self.max_tokens = max_tokens
+        self.temperature = temperature
+        self._session = requests.Session()
+        self.available = self._detect_availability()
+
+    # ------------------------------------------------------------------
+    # Public helpers
+    # ------------------------------------------------------------------
+    def generate_prediction(self, payload: Dict[str, Any]) -> Dict[str, Any]:
+        self._ensure_ready()
+        scenario = json.dumps(payload, ensure_ascii=False, indent=2)
+        messages = [
+            self._system(
+                "You are an elite business strategist. Return a JSON object with fields: "
+                "summary, growth_outlook (0-100), confidence (0-100), key_initiatives (array of strings), "
+                "insight_cards (array of {title, detail}), and chart {labels: string[], series: number[]} "
+                "representing projected impact areas. Use concise business language."
+            ),
+            self._user(f"Scenario data:```json\n{scenario}\n```"),
+        ]
+        return self._respond(messages)
+
+    def assess_risk(self, payload: Dict[str, Any]) -> Dict[str, Any]:
+        self._ensure_ready()
+        scenario = json.dumps(payload, ensure_ascii=False, indent=2)
+        messages = [
+            self._system(
+                "You are a chief risk officer. Return JSON with fields: risk_summary, overall_score (0-100), "
+                "risk_matrix (array of {name, likelihood, impact, mitigation, detail}), early_warnings (string array), "
+                "and heatmap {labels: string[], series: number[]} for visualization."
+            ),
+            self._user(f"Business context:```json\n{scenario}\n```"),
+        ]
+        return self._respond(messages)
+
+    def provide_advice(self, payload: Dict[str, Any]) -> Dict[str, Any]:
+        self._ensure_ready()
+        context = json.dumps(payload, ensure_ascii=False, indent=2)
+        messages = [
+            self._system(
+                "You are an executive advisor. Reply as JSON with fields: recommendation, execution_steps (array), "
+                "follow_up_questions (array), and sentiment (supportive | neutral | cautionary)."
+            ),
+            self._user(f"Context for advice:```json\n{context}\n```"),
+        ]
+        return self._respond(messages)
+
+    def chat(self, conversation: Iterable[Dict[str, str]], context: Optional[str] = None) -> Dict[str, Any]:
+        self._ensure_ready()
+        messages = [self._system("You are a CRM co-pilot offering crisp, actionable insights.")]
+        if context:
+            messages.append(self._user(f"Reference context: {context}"))
+        for turn in conversation:
+            role = turn.get("role") or "user"
+            content = turn.get("content") or ""
+            messages.append({"role": role, "content": content})
+        response = self._raw_request(messages)
+        answer = response.get("choices", [{}])[0].get("message", {}).get("content", "")
+        return {"reply": answer}
+
+    def compare_documents(self, left: Dict[str, str], right: Dict[str, str]) -> Dict[str, Any]:
+        self._ensure_ready()
+        left_excerpt = left.get("text", "")[:6000]
+        right_excerpt = right.get("text", "")[:6000]
+        messages = [
+            self._system(
+                "You are a principal consultant performing deep document comparison. "
+                "Return JSON with sections: synopsis (string), alignment (array of {theme, assessment}), "
+                "divergence (array of {issue, severity, detail}), recommendations (array), "
+                "and comparison_score (0-100)."
+            ),
+            self._user(
+                f"Document A ({left.get('name')}):\n```\n{left_excerpt}\n```\n\n"
+                f"Document B ({right.get('name')}):\n```\n{right_excerpt}\n```"
+            ),
+        ]
+        return self._respond(messages)
+
+    def summarize_document(self, document: Dict[str, str]) -> Dict[str, Any]:
+        self._ensure_ready()
+        snippet = document.get("text", "")[:6000]
+        messages = [
+            self._system(
+                "Summarize the document for CRM stakeholders. Respond as JSON with fields: summary, tags (array), "
+                "data_quality (0-100), and opportunities (array)."
+            ),
+            self._user(
+                f"Document {document.get('name')} excerpt:\n```\n{snippet}\n```"
+            ),
+        ]
+        return self._respond(messages)
+
+    # ------------------------------------------------------------------
+    # Internal helpers
+    # ------------------------------------------------------------------
+    def _respond(self, messages: List[Dict[str, str]]) -> Dict[str, Any]:
+        response = self._raw_request(messages)
+        content = response.get("choices", [{}])[0].get("message", {}).get("content", "")
+        return self._parse_json(content)
+
+    def _raw_request(self, messages: List[Dict[str, str]]) -> Dict[str, Any]:
+        payload = {
+            "model": self.model,
+            "messages": messages,
+            "stream": False,
+            "options": {
+                "temperature": self.temperature,
+                "num_ctx": self.context,
+                "num_predict": self.max_tokens,
+            },
+        }
+        url = f"{self.host}/api/chat"
+        try:
+            response = self._session.post(url, json=payload, timeout=120)
+            response.raise_for_status()
+        except requests.RequestException as exc:  # pragma: no cover - network
+            raise InsightServiceError(
+                "Unable to contact the local Ollama service. Ensure Ollama is running and the model is pulled."
+            ) from exc
+
+        try:
+            data = response.json()
+        except ValueError as exc:  # pragma: no cover - invalid JSON from runtime
+            raise InsightServiceError("The local model returned an invalid response. Review your Ollama logs.") from exc
+
+        if isinstance(data, dict) and data.get("error"):
+            raise InsightServiceError(f"Ollama error: {data['error']}")
+
+        message = (data.get("message") or {}).get("content", "")
+        return {"choices": [{"message": {"content": message}}]}
+
+    def _ensure_ready(self) -> None:
+        if not self.available:
+            raise InsightServiceError(
+                "Local inference is not configured. Install and start Ollama, set OLLAMA_HOST, and specify LLM_MODEL in your .env."
+            )
+
+    def _detect_availability(self) -> bool:
+        if not self.model:
+            return False
+        try:
+            response = self._session.get(f"{self.host}/api/tags", timeout=5)
+            response.raise_for_status()
+            payload = response.json()
+        except requests.RequestException:
+            return False
+        except ValueError:
+            return False
+
+        models = payload.get("models", []) if isinstance(payload, dict) else []
+        if not models:
+            return True
+        return any(
+            self.model == entry.get("name") or self.model == entry.get("model")
+            for entry in models
+            if isinstance(entry, dict)
+        )
+
+    @staticmethod
+    def _system(content: str) -> Dict[str, str]:
+        return {"role": "system", "content": content}
+
+    @staticmethod
+    def _user(content: str) -> Dict[str, str]:
+        return {"role": "user", "content": content}
+
+    @staticmethod
+    def _parse_json(payload: str) -> Dict[str, Any]:
+        try:
+            return json.loads(payload)
+        except json.JSONDecodeError:
+            match = re.search(r"\{[\s\S]*\}", payload)
+            if match:
+                try:
+                    return json.loads(match.group(0))
+                except json.JSONDecodeError as exc:
+                    raise InsightServiceError(
+                        "The AI response was not valid JSON. Adjust your prompt or retry."
+                    ) from exc
+            raise InsightServiceError("The AI response was not valid JSON. Adjust your prompt or retry.")
diff --git a/app/static/css/styles.css b/app/static/css/styles.css
new file mode 100644
index 0000000..2f91354
--- /dev/null
+++ b/app/static/css/styles.css
@@ -0,0 +1,1181 @@
+:root {
+  --font-sans: "Inter", "Segoe UI", "Manrope", system-ui, -apple-system, BlinkMacSystemFont, sans-serif;
+  --bg-primary: #050513;
+  --bg-elevated: rgba(18, 21, 45, 0.72);
+  --bg-elevated-strong: rgba(30, 33, 64, 0.84);
+  --bg-panel: rgba(18, 22, 48, 0.82);
+  --bg-panel-soft: rgba(25, 29, 62, 0.7);
+  --bg-glow: linear-gradient(135deg, rgba(94, 147, 255, 0.22), rgba(120, 72, 255, 0.18));
+  --border-subtle: rgba(255, 255, 255, 0.08);
+  --border-strong: rgba(94, 147, 255, 0.45);
+  --text-primary: #f8f9ff;
+  --text-secondary: rgba(248, 249, 255, 0.72);
+  --text-muted: rgba(248, 249, 255, 0.48);
+  --accent: #6f8bff;
+  --accent-strong: #4dd0ff;
+  --accent-hot: #ff7ab8;
+  --danger: #ff6b6b;
+  --warning: #ffc861;
+  --success: #58f29a;
+  --shadow-soft: 0 25px 60px -35px rgba(0, 0, 0, 0.65);
+  --radius-xl: 34px;
+  --radius-lg: 26px;
+  --radius-md: 18px;
+  --radius-sm: 12px;
+  color-scheme: dark;
+}
+
+* {
+  box-sizing: border-box;
+}
+
+body {
+  min-height: 100vh;
+  margin: 0;
+  font-family: var(--font-sans);
+  background:
+    radial-gradient(circle at 12% 18%, rgba(109, 173, 255, 0.22), transparent 55%),
+    radial-gradient(circle at 85% 12%, rgba(255, 112, 173, 0.18), transparent 65%),
+    radial-gradient(circle at 65% 85%, rgba(77, 208, 255, 0.14), transparent 60%),
+    var(--bg-primary);
+  color: var(--text-primary);
+  line-height: 1.6;
+  -webkit-font-smoothing: antialiased;
+  transition: background 0.3s ease;
+}
+
+body.focus-mode {
+  background: radial-gradient(circle at 50% 0%, rgba(36, 53, 120, 0.3), transparent 70%), #020209;
+}
+
+.background-veil {
+  position: fixed;
+  inset: 0;
+  z-index: -2;
+  pointer-events: none;
+}
+
+.background-veil::before,
+.background-veil::after {
+  content: "";
+  position: absolute;
+  border-radius: 50%;
+  filter: blur(140px);
+  opacity: 0.65;
+}
+
+.background-veil::before {
+  width: 460px;
+  height: 460px;
+  top: -140px;
+  left: -110px;
+  background: radial-gradient(circle at 30% 30%, rgba(125, 151, 255, 0.45), transparent 70%);
+}
+
+.background-veil::after {
+  width: 520px;
+  height: 520px;
+  bottom: -220px;
+  right: -160px;
+  background: radial-gradient(circle at 50% 50%, rgba(255, 120, 184, 0.32), transparent 65%);
+}
+
+.container {
+  width: min(1280px, 92vw);
+  margin: 0 auto;
+}
+
+.shell-header {
+  position: sticky;
+  top: 0;
+  z-index: 20;
+  background: rgba(6, 7, 18, 0.7);
+  backdrop-filter: blur(24px);
+  border-bottom: 1px solid rgba(255, 255, 255, 0.04);
+}
+
+.shell-header__inner {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  padding: 1.1rem 0;
+}
+
+.brand-mark {
+  display: flex;
+  align-items: center;
+  gap: 0.9rem;
+  text-decoration: none;
+  color: var(--text-primary);
+  font-weight: 700;
+  letter-spacing: 0.04em;
+  text-transform: uppercase;
+}
+
+.brand-icon {
+  width: 38px;
+  height: 38px;
+  display: grid;
+  place-items: center;
+  border-radius: 14px;
+  background: linear-gradient(135deg, rgba(112, 142, 255, 0.75), rgba(77, 208, 255, 0.75));
+  box-shadow: inset 0 0 0 1px rgba(255, 255, 255, 0.4);
+  font-weight: 800;
+}
+
+.shell-header__actions {
+  display: flex;
+  align-items: center;
+  gap: 0.75rem;
+}
+
+.nav-toggle {
+  display: none;
+  width: 44px;
+  height: 44px;
+  border-radius: 14px;
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  background: rgba(18, 23, 46, 0.65);
+  color: var(--text-primary);
+  font-size: 1.2rem;
+}
+
+.shell-nav {
+  display: flex;
+  gap: 0.4rem;
+  padding-bottom: 1.1rem;
+}
+
+.shell-nav__link {
+  padding: 0.55rem 1.25rem;
+  border-radius: 999px;
+  text-decoration: none;
+  font-size: 0.9rem;
+  letter-spacing: 0.08em;
+  text-transform: uppercase;
+  color: var(--text-secondary);
+  background: transparent;
+  border: 1px solid transparent;
+  transition: background 0.3s ease, border-color 0.3s ease, color 0.3s ease;
+}
+
+.shell-nav__link:hover {
+  color: var(--text-primary);
+  border-color: rgba(255, 255, 255, 0.12);
+  background: rgba(111, 139, 255, 0.12);
+}
+
+.shell-nav__link.is-active {
+  color: var(--text-primary);
+  border-color: rgba(255, 255, 255, 0.32);
+  background: linear-gradient(135deg, rgba(111, 139, 255, 0.26), rgba(77, 208, 255, 0.26));
+}
+
+.status-strip {
+  padding: 1.75rem 0 1rem;
+}
+
+.status-strip__inner {
+  display: flex;
+  flex-direction: column;
+  gap: 1.5rem;
+}
+
+.status-grid {
+  display: grid;
+  grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
+  gap: 1rem;
+}
+
+.status-card {
+  position: relative;
+  padding: 1.35rem 1.5rem;
+  border-radius: var(--radius-lg);
+  background: var(--bg-elevated);
+  border: 1px solid rgba(255, 255, 255, 0.06);
+  box-shadow: var(--shadow-soft);
+}
+
+.status-card__label {
+  display: block;
+  font-size: 0.75rem;
+  letter-spacing: 0.32em;
+  text-transform: uppercase;
+  color: var(--text-muted);
+}
+
+.status-card__value {
+  margin-top: 0.75rem;
+  font-size: 1.1rem;
+  font-weight: 600;
+  display: inline-flex;
+  align-items: center;
+  gap: 0.65rem;
+}
+
+.status-card__value .dot {
+  width: 11px;
+  height: 11px;
+  border-radius: 50%;
+  background: rgba(255, 255, 255, 0.5);
+  box-shadow: 0 0 0 4px rgba(255, 255, 255, 0.08);
+}
+
+.status-card__value.online .dot {
+  background: var(--success);
+}
+
+.status-card__value.offline .dot {
+  background: var(--warning);
+}
+
+.status-card__hint {
+  margin-top: 0.45rem;
+  font-size: 0.78rem;
+  color: var(--text-muted);
+}
+
+.command-strip {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 0.75rem;
+}
+
+.command-btn {
+  border: 1px solid rgba(255, 255, 255, 0.1);
+  color: var(--text-primary);
+  background: rgba(13, 16, 34, 0.72);
+  padding: 0.6rem 1.5rem;
+  border-radius: 999px;
+  font-size: 0.78rem;
+  letter-spacing: 0.18em;
+  text-transform: uppercase;
+  transition: transform 0.3s ease, background 0.3s ease;
+}
+
+.command-btn:hover {
+  background: rgba(111, 139, 255, 0.2);
+  transform: translateY(-2px);
+}
+
+.disabled {
+  opacity: 0.5;
+  pointer-events: none;
+}
+
+.page-shell {
+  padding: 1rem 0 4rem;
+}
+
+.flash-stack {
+  display: grid;
+  gap: 0.8rem;
+  margin-bottom: 2rem;
+}
+
+.flash-message {
+  padding: 1rem 1.25rem;
+  border-radius: var(--radius-md);
+  background: rgba(255, 255, 255, 0.08);
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  font-size: 0.95rem;
+}
+
+.flash-message.success {
+  border-color: rgba(120, 255, 190, 0.4);
+  background: rgba(120, 255, 190, 0.12);
+}
+
+.flash-message.danger {
+  border-color: rgba(255, 120, 160, 0.4);
+  background: rgba(255, 120, 160, 0.12);
+}
+
+code,
+kbd {
+  background: rgba(255, 255, 255, 0.08);
+  border-radius: 6px;
+  padding: 0.15rem 0.4rem;
+  font-family: var(--font-sans);
+  font-size: 0.75rem;
+}
+
+.panel {
+  position: relative;
+  border-radius: var(--radius-xl);
+  background: var(--bg-panel);
+  padding: clamp(1.75rem, 1.6vw + 1.4rem, 3rem);
+  border: 1px solid rgba(255, 255, 255, 0.07);
+  box-shadow: var(--shadow-soft);
+  backdrop-filter: blur(28px);
+}
+
+.panel::before {
+  content: "";
+  position: absolute;
+  inset: 0;
+  border-radius: inherit;
+  border: 1px solid rgba(255, 255, 255, 0.05);
+  pointer-events: none;
+  mix-blend-mode: screen;
+}
+
+.panel--soft {
+  background: var(--bg-panel-soft);
+}
+
+.hero-grid {
+  display: grid;
+  gap: clamp(2rem, 2vw + 1rem, 3.5rem);
+  grid-template-columns: repeat(auto-fit, minmax(280px, 1fr));
+}
+
+.hero-eyebrow {
+  display: inline-flex;
+  align-items: center;
+  gap: 0.6rem;
+  padding: 0.4rem 1.2rem;
+  border-radius: 999px;
+  font-size: 0.75rem;
+  letter-spacing: 0.32em;
+  text-transform: uppercase;
+  color: var(--text-primary);
+  background: linear-gradient(135deg, rgba(111, 139, 255, 0.24), rgba(77, 208, 255, 0.2));
+}
+
+.hero-eyebrow .dot {
+  width: 8px;
+  height: 8px;
+  border-radius: 50%;
+  background: var(--success);
+}
+
+.display-xl {
+  font-size: clamp(2.8rem, 2.4vw + 2.2rem, 4.2rem);
+  line-height: 1.08;
+  font-weight: 700;
+  margin: 1.1rem 0 1.5rem;
+  background: linear-gradient(135deg, #fff, rgba(180, 207, 255, 0.9));
+  -webkit-background-clip: text;
+  color: transparent;
+}
+
+.display-md {
+  font-size: clamp(1.8rem, 1.4vw + 1.3rem, 2.6rem);
+  line-height: 1.2;
+  font-weight: 600;
+  margin: 0.9rem 0 1.1rem;
+}
+
+.text-muted {
+  color: var(--text-secondary);
+}
+
+.text-dimmed {
+  color: var(--text-muted);
+}
+
+.kpi-grid {
+  display: grid;
+  gap: 1.2rem;
+}
+
+.kpi-card {
+  border-radius: var(--radius-lg);
+  background: rgba(12, 16, 36, 0.75);
+  border: 1px solid rgba(255, 255, 255, 0.07);
+  padding: 1.4rem 1.6rem;
+  display: grid;
+  gap: 0.6rem;
+}
+
+.kpi-card.primary {
+  background: linear-gradient(135deg, rgba(104, 135, 255, 0.35), rgba(77, 205, 255, 0.2));
+  border-color: rgba(120, 180, 255, 0.5);
+}
+
+.kpi-label {
+  text-transform: uppercase;
+  letter-spacing: 0.26em;
+  font-size: 0.72rem;
+  color: var(--text-muted);
+}
+
+.kpi-value {
+  font-size: 1.9rem;
+  font-weight: 700;
+}
+
+.hero-actions {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 0.85rem;
+  margin-top: 2rem;
+}
+
+.btn {
+  display: inline-flex;
+  align-items: center;
+  justify-content: center;
+  gap: 0.6rem;
+  border-radius: 16px;
+  border: none;
+  padding: 0.85rem 1.65rem;
+  font-size: 0.95rem;
+  font-weight: 600;
+  letter-spacing: 0.04em;
+  color: var(--text-primary);
+  background: rgba(16, 18, 36, 0.8);
+  border: 1px solid rgba(255, 255, 255, 0.06);
+  cursor: pointer;
+  transition: transform 0.25s ease, box-shadow 0.25s ease, background 0.25s ease;
+}
+
+.btn:hover {
+  transform: translateY(-2px);
+  box-shadow: 0 18px 32px -24px rgba(111, 139, 255, 0.8);
+}
+
+.btn--primary {
+  background: linear-gradient(135deg, rgba(111, 139, 255, 0.85), rgba(77, 208, 255, 0.85));
+  border-color: rgba(255, 255, 255, 0.24);
+}
+
+.btn--secondary {
+  background: rgba(12, 18, 44, 0.8);
+}
+
+.btn--outline {
+  background: transparent;
+  border-color: rgba(255, 255, 255, 0.25);
+}
+
+.btn--ghost {
+  background: rgba(255, 255, 255, 0.08);
+  border-color: transparent;
+}
+
+.btn--danger {
+  background: linear-gradient(135deg, rgba(255, 122, 152, 0.85), rgba(255, 107, 107, 0.85));
+}
+
+.btn--full {
+  width: 100%;
+}
+
+.btn--small {
+  padding: 0.55rem 1.1rem;
+  font-size: 0.8rem;
+}
+
+.scroll-link {
+  display: inline-flex;
+  align-items: center;
+  gap: 0.55rem;
+  padding: 0.5rem 1rem;
+  border-radius: 999px;
+  background: rgba(255, 255, 255, 0.08);
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  color: var(--text-primary);
+  text-decoration: none;
+}
+
+.marquee {
+  overflow: hidden;
+  position: relative;
+  border-radius: var(--radius-md);
+  border: 1px solid rgba(255, 255, 255, 0.07);
+  background: rgba(10, 12, 28, 0.72);
+  margin-top: 1.8rem;
+}
+
+.marquee-track {
+  display: inline-flex;
+  gap: 2.5rem;
+  padding: 0.9rem 1.4rem;
+  animation: marquee 24s linear infinite;
+}
+
+.marquee-item {
+  font-size: 0.9rem;
+  color: var(--text-secondary);
+}
+
+@keyframes marquee {
+  0% {
+    transform: translateX(0);
+  }
+  100% {
+    transform: translateX(-50%);
+  }
+}
+
+.spotlight-grid {
+  display: grid;
+  gap: 1rem;
+  margin-top: 1.5rem;
+}
+
+.spotlight-card {
+  padding: 1rem 1.25rem;
+  border-radius: var(--radius-md);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  background: rgba(14, 16, 34, 0.72);
+  display: grid;
+  gap: 0.35rem;
+}
+
+.spotlight-card header {
+  display: flex;
+  justify-content: space-between;
+  font-size: 0.78rem;
+  color: var(--text-muted);
+  letter-spacing: 0.18em;
+  text-transform: uppercase;
+}
+
+.badge {
+  display: inline-flex;
+  align-items: center;
+  padding: 0.35rem 0.85rem;
+  border-radius: 999px;
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  font-size: 0.75rem;
+  letter-spacing: 0.08em;
+  text-transform: uppercase;
+  color: var(--text-secondary);
+  background: rgba(255, 255, 255, 0.06);
+}
+
+.badge--muted {
+  background: rgba(255, 255, 255, 0.12);
+}
+
+.coverage-bars {
+  display: grid;
+  gap: 0.8rem;
+  margin-top: 1.3rem;
+}
+
+.coverage-bar {
+  display: grid;
+  gap: 0.4rem;
+}
+
+.coverage-bar .bar {
+  position: relative;
+  border-radius: 999px;
+  background: rgba(255, 255, 255, 0.08);
+  overflow: hidden;
+  height: 8px;
+}
+
+.coverage-bar .bar span {
+  position: absolute;
+  inset: 0;
+  background: linear-gradient(135deg, rgba(111, 139, 255, 0.65), rgba(77, 208, 255, 0.65));
+}
+
+.feature-grid {
+  display: grid;
+  gap: 1.75rem;
+  grid-template-columns: repeat(auto-fit, minmax(240px, 1fr));
+  margin-top: 3rem;
+}
+
+.feature-card {
+  border-radius: var(--radius-lg);
+  background: rgba(12, 16, 34, 0.72);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  padding: 1.6rem;
+  display: grid;
+  gap: 0.9rem;
+  transition: transform 0.3s ease, background 0.3s ease;
+}
+
+.feature-card:hover {
+  transform: translateY(-6px);
+  background: rgba(22, 28, 56, 0.85);
+}
+
+.feature-icon {
+  font-size: 1.75rem;
+}
+
+.section-heading {
+  display: flex;
+  flex-wrap: wrap;
+  align-items: center;
+  justify-content: space-between;
+  gap: 1rem;
+  margin-bottom: 1.8rem;
+}
+
+.section-heading .eyebrow {
+  font-size: 0.75rem;
+  letter-spacing: 0.32em;
+  text-transform: uppercase;
+  color: var(--text-muted);
+}
+
+.eyebrow {
+  font-size: 0.74rem;
+  letter-spacing: 0.32em;
+  text-transform: uppercase;
+  color: var(--text-muted);
+  margin: 0;
+}
+
+.timeline {
+  list-style: none;
+  margin: 0;
+  padding: 0;
+  display: grid;
+  gap: 1.1rem;
+}
+
+.timeline-item {
+  position: relative;
+  padding-left: 1.75rem;
+}
+
+.timeline-item::before {
+  content: "";
+  position: absolute;
+  left: 0.4rem;
+  top: 0.35rem;
+  width: 10px;
+  height: 10px;
+  border-radius: 50%;
+  background: var(--accent);
+  box-shadow: 0 0 12px rgba(111, 139, 255, 0.5);
+}
+
+.timeline-item::after {
+  content: "";
+  position: absolute;
+  left: 0.85rem;
+  top: 1.2rem;
+  bottom: -1.2rem;
+  width: 1px;
+  background: rgba(255, 255, 255, 0.18);
+}
+
+.timeline-item:last-child::after {
+  display: none;
+}
+
+.story-grid {
+  display: grid;
+  gap: 1.25rem;
+  grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
+}
+
+.story-card {
+  border-radius: var(--radius-lg);
+  padding: 1.35rem;
+  background: rgba(12, 16, 38, 0.75);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  display: grid;
+  gap: 0.6rem;
+}
+
+.story-card.highlight {
+  background: linear-gradient(135deg, rgba(111, 139, 255, 0.28), rgba(77, 208, 255, 0.24));
+  border-color: rgba(255, 255, 255, 0.16);
+}
+
+.story-card ul {
+  padding-left: 1rem;
+  margin: 0;
+  color: var(--text-secondary);
+}
+
+.catalog-grid {
+  display: grid;
+  gap: 1.1rem;
+}
+
+.catalog-card {
+  border-radius: var(--radius-lg);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  background: rgba(14, 16, 36, 0.78);
+  padding: 1.1rem 1.35rem;
+  display: grid;
+  gap: 0.55rem;
+}
+
+.catalog-card header {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  font-size: 0.75rem;
+  letter-spacing: 0.22em;
+  text-transform: uppercase;
+  color: var(--text-muted);
+}
+
+.catalog-actions {
+  margin-top: 0.6rem;
+}
+
+.catalog-actions .btn {
+  width: fit-content;
+  padding: 0.5rem 1.1rem;
+  font-size: 0.78rem;
+}
+
+.form-grid {
+  display: grid;
+  gap: 1.1rem;
+}
+
+.form-field {
+  display: grid;
+  gap: 0.45rem;
+}
+
+.form-label {
+  font-size: 0.85rem;
+  text-transform: uppercase;
+  letter-spacing: 0.22em;
+  color: var(--text-muted);
+}
+
+.form-control,
+.form-select,
+textarea {
+  width: 100%;
+  padding: 0.85rem 1rem;
+  border-radius: var(--radius-md);
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  background: rgba(10, 12, 28, 0.72);
+  color: var(--text-primary);
+  font-size: 0.95rem;
+  font-family: var(--font-sans);
+}
+
+.form-control:focus,
+.form-select:focus,
+textarea:focus {
+  outline: none;
+  border-color: rgba(111, 139, 255, 0.6);
+  box-shadow: 0 0 0 3px rgba(111, 139, 255, 0.25);
+}
+
+.form-select[multiple] {
+  min-height: 180px;
+}
+
+.template-tray {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 0.6rem;
+  margin-top: 1.2rem;
+}
+
+.template-chip {
+  padding: 0.45rem 1.2rem;
+  border-radius: 999px;
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  background: rgba(255, 255, 255, 0.08);
+  color: var(--text-primary);
+  font-size: 0.8rem;
+  cursor: pointer;
+  transition: transform 0.25s ease;
+}
+
+.template-chip:hover {
+  transform: translateY(-2px);
+}
+
+.metric-board {
+  display: grid;
+  gap: 1rem;
+  grid-template-columns: repeat(auto-fit, minmax(180px, 1fr));
+}
+
+.metric-card {
+  border-radius: var(--radius-lg);
+  background: rgba(12, 16, 36, 0.78);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  padding: 1rem 1.2rem;
+  display: grid;
+  gap: 0.4rem;
+}
+
+.metric-card .label {
+  font-size: 0.72rem;
+  letter-spacing: 0.32em;
+  text-transform: uppercase;
+  color: var(--text-muted);
+}
+
+.metric-card .value {
+  font-size: 1.6rem;
+  font-weight: 600;
+}
+
+.metric-dial {
+  position: relative;
+  width: 130px;
+  height: 130px;
+  border-radius: 50%;
+  display: grid;
+  place-items: center;
+  background: conic-gradient(
+    var(--accent) calc(var(--score, 0) * 1%),
+    rgba(255, 255, 255, 0.08) 0
+  );
+}
+
+.metric-dial::after {
+  content: "";
+  position: absolute;
+  inset: 12px;
+  border-radius: 50%;
+  background: rgba(10, 12, 28, 0.9);
+}
+
+.metric-dial .value {
+  position: relative;
+  font-size: 2rem;
+  font-weight: 700;
+}
+
+.metric-dial[data-score] .value::after {
+  content: "%";
+  font-size: 1rem;
+  font-weight: 600;
+  margin-left: 2px;
+}
+
+.inline-alert {
+  border-radius: var(--radius-md);
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  padding: 0.85rem 1rem;
+  font-size: 0.9rem;
+  background: rgba(255, 255, 255, 0.06);
+}
+
+.inline-alert.error {
+  border-color: rgba(255, 136, 156, 0.55);
+  background: rgba(255, 136, 156, 0.18);
+}
+
+.inline-alert.warning {
+  border-color: rgba(255, 210, 128, 0.45);
+  background: rgba(255, 210, 128, 0.18);
+}
+
+.loading-spinner {
+  width: 48px;
+  height: 48px;
+  border-radius: 50%;
+  border: 4px solid rgba(255, 255, 255, 0.12);
+  border-top-color: var(--accent);
+  animation: spin 1s linear infinite;
+  margin: 0 auto;
+}
+
+@keyframes spin {
+  to {
+    transform: rotate(360deg);
+  }
+}
+
+.list-stack {
+  list-style: none;
+  margin: 0;
+  padding: 0;
+  display: grid;
+  gap: 0.65rem;
+}
+
+.list-stack__item {
+  padding: 0.75rem 0.95rem;
+  border-radius: var(--radius-md);
+  background: rgba(255, 255, 255, 0.06);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  font-size: 0.9rem;
+}
+
+.insight-pillboard {
+  display: grid;
+  gap: 1rem;
+  grid-template-columns: repeat(auto-fit, minmax(210px, 1fr));
+  margin-top: 1.5rem;
+}
+
+.insight-pill {
+  border-radius: var(--radius-lg);
+  padding: 1.15rem 1.3rem;
+  background: rgba(12, 16, 36, 0.78);
+  border: 1px solid rgba(255, 255, 255, 0.1);
+  display: grid;
+  gap: 0.45rem;
+}
+
+.ai-panel {
+  border-radius: var(--radius-lg);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  background: rgba(10, 12, 28, 0.72);
+  padding: 1.3rem 1.5rem;
+  display: grid;
+  gap: 0.8rem;
+}
+
+.summary-grid {
+  display: grid;
+  gap: 1.2rem;
+}
+
+.summary-block {
+  display: grid;
+  gap: 0.6rem;
+}
+
+.summary-block p {
+  margin: 0;
+}
+
+.summary-block strong {
+  font-size: 1rem;
+}
+
+.tag-cloud {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 0.5rem;
+}
+
+.quality-dial {
+  display: grid;
+  place-items: center;
+  width: 160px;
+  height: 160px;
+  margin: 1.5rem auto 0;
+  position: relative;
+  --score: 0;
+  background: conic-gradient(var(--accent-strong) calc(var(--score) * 1%), rgba(255, 255, 255, 0.08) 0);
+  border-radius: 50%;
+}
+
+.quality-dial::after {
+  content: "";
+  position: absolute;
+  inset: 16px;
+  border-radius: 50%;
+  background: rgba(6, 8, 20, 0.92);
+}
+
+.quality-dial .value {
+  position: relative;
+  font-size: 2.2rem;
+  font-weight: 700;
+}
+
+.chat-wrapper {
+  display: grid;
+  gap: 1.5rem;
+  grid-template-columns: minmax(0, 3fr) minmax(0, 2fr);
+}
+
+.chat-log {
+  border-radius: var(--radius-lg);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  background: rgba(10, 12, 28, 0.72);
+  padding: 1.25rem;
+  display: grid;
+  gap: 1rem;
+  max-height: 540px;
+  overflow-y: auto;
+}
+
+.chat-message {
+  border-radius: var(--radius-md);
+  padding: 0.9rem 1rem;
+  background: rgba(255, 255, 255, 0.06);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+}
+
+.chat-message.me {
+  background: rgba(255, 255, 255, 0.12);
+  border-color: rgba(255, 255, 255, 0.16);
+}
+
+.chat-message .label {
+  display: block;
+  font-size: 0.75rem;
+  letter-spacing: 0.28em;
+  text-transform: uppercase;
+  margin-bottom: 0.4rem;
+  color: var(--text-muted);
+}
+
+.chat-console {
+  display: grid;
+  gap: 1rem;
+}
+
+.chat-console textarea {
+  min-height: 120px;
+}
+
+.prompt-chip {
+  padding: 0.45rem 1rem;
+  border-radius: 999px;
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  background: rgba(255, 255, 255, 0.08);
+  font-size: 0.78rem;
+  cursor: pointer;
+}
+
+.prompt-chip:hover {
+  transform: translateY(-2px);
+}
+
+.comparison-grid {
+  display: grid;
+  gap: 1.2rem;
+  grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
+  margin-top: 1.5rem;
+}
+
+.difference-group {
+  border-radius: var(--radius-lg);
+  border: 1px solid rgba(255, 255, 255, 0.08);
+  background: rgba(12, 16, 36, 0.78);
+  padding: 1.2rem 1.4rem;
+  display: grid;
+  gap: 0.65rem;
+}
+
+.difference-group h4 {
+  margin: 0;
+  text-transform: uppercase;
+  letter-spacing: 0.22em;
+  font-size: 0.78rem;
+  color: var(--text-muted);
+}
+
+.difference-group ul {
+  margin: 0;
+  padding-left: 1rem;
+  color: var(--text-secondary);
+}
+
+.playbook-drawer {
+  position: fixed;
+  inset: 0;
+  background: rgba(3, 5, 16, 0.75);
+  backdrop-filter: blur(14px);
+  display: grid;
+  place-items: center;
+  opacity: 0;
+  pointer-events: none;
+  transition: opacity 0.35s ease;
+  z-index: 50;
+}
+
+.playbook-drawer.open {
+  opacity: 1;
+  pointer-events: auto;
+}
+
+.playbook-panel {
+  width: min(520px, 92vw);
+  border-radius: var(--radius-xl);
+  background: rgba(14, 16, 36, 0.92);
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  padding: 2.4rem;
+  box-shadow: var(--shadow-soft);
+  display: grid;
+  gap: 1.6rem;
+}
+
+.playbook-panel header {
+  display: flex;
+  justify-content: space-between;
+  align-items: flex-start;
+}
+
+.icon-btn {
+  width: 42px;
+  height: 42px;
+  border-radius: 14px;
+  border: 1px solid rgba(255, 255, 255, 0.14);
+  background: rgba(16, 18, 34, 0.75);
+  color: var(--text-primary);
+  font-size: 1.1rem;
+  display: grid;
+  place-items: center;
+  cursor: pointer;
+}
+
+.icon-btn:hover {
+  background: rgba(111, 139, 255, 0.2);
+}
+
+.aurora-toast {
+  position: fixed;
+  bottom: 2rem;
+  right: 2rem;
+  padding: 0.85rem 1.1rem;
+  border-radius: var(--radius-md);
+  background: rgba(12, 16, 38, 0.95);
+  border: 1px solid rgba(255, 255, 255, 0.12);
+  opacity: 0;
+  transform: translateY(12px);
+  pointer-events: none;
+  transition: opacity 0.3s ease, transform 0.3s ease;
+  z-index: 60;
+}
+
+.aurora-toast.visible {
+  opacity: 1;
+  transform: translateY(0);
+}
+
+@media (max-width: 960px) {
+  .shell-nav {
+    display: none;
+    flex-direction: column;
+    gap: 0.6rem;
+    padding: 0 0 1.4rem;
+  }
+
+  .shell-nav.open {
+    display: flex;
+  }
+
+  .nav-toggle {
+    display: grid;
+  }
+
+  .chat-wrapper {
+    grid-template-columns: 1fr;
+  }
+}
+
+@media (max-width: 720px) {
+  .command-strip {
+    flex-direction: column;
+    align-items: stretch;
+  }
+
+  .hero-actions {
+    flex-direction: column;
+    align-items: stretch;
+  }
+
+  .metric-board,
+  .feature-grid,
+  .story-grid {
+    grid-template-columns: 1fr;
+  }
+
+  .hero-grid {
+    grid-template-columns: 1fr;
+  }
+}
diff --git a/app/static/js/advisor.js b/app/static/js/advisor.js
new file mode 100644
index 0000000..e26f44d
--- /dev/null
+++ b/app/static/js/advisor.js
@@ -0,0 +1,104 @@
+const chatLog = document.getElementById('chatLog');
+const chatForm = document.getElementById('chatForm');
+const chatPrompt = document.getElementById('chatPrompt');
+const chatStatus = document.getElementById('chatStatus');
+const advisorContext = document.getElementById('advisorContext');
+const advisorDocuments = document.getElementById('advisorDocuments');
+const loadContextBtn = document.getElementById('loadContext');
+const advisorContextSummary = document.getElementById('advisorContextSummary');
+const promptChips = document.querySelectorAll('.prompt-chip');
+const conversation = [];
+
+function renderMessage(role, content) {
+  const wrapper = document.createElement('div');
+  wrapper.classList.add('chat-message', role === 'user' ? 'me' : 'aurora');
+  wrapper.innerHTML = `
+    <span class="label">${role === 'user' ? 'You' : 'Aurora'}</span>
+    <div>${content}</div>
+  `;
+  chatLog.appendChild(wrapper);
+  chatLog.scrollTop = chatLog.scrollHeight;
+}
+
+async function sendMessage(message) {
+  chatStatus.textContent = 'Thinking…';
+  conversation.push({ role: 'user', content: message });
+  renderMessage('user', message);
+  chatPrompt.value = '';
+  try {
+    const payload = {
+      conversation,
+      context: advisorContext.value,
+    };
+    const response = await fetch('/api/chat', {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify(payload),
+    });
+    const data = await response.json();
+    if (!response.ok) throw new Error(data.error || 'Unable to complete chat request');
+    const reply = data.reply || 'No response received.';
+    conversation.push({ role: 'assistant', content: reply });
+    renderMessage('assistant', reply);
+    chatStatus.textContent = 'Ready';
+  } catch (error) {
+    chatStatus.textContent = 'Error';
+    renderMessage('assistant', `<div class="inline-alert error">${error.message}</div>`);
+  }
+}
+
+async function loadSummaries() {
+  advisorContextSummary.innerHTML = '<div class="loading-spinner" aria-hidden="true"></div>';
+  const selected = Array.from(advisorDocuments?.selectedOptions || []).map((option) => option.value);
+  if (!selected.length) {
+    advisorContextSummary.innerHTML = '<p class="text-muted">Select documents to load quick summaries.</p>';
+    return;
+  }
+  try {
+    const results = [];
+    for (const filename of selected) {
+      const response = await fetch('/api/data/summarize', {
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({ filename }),
+      });
+      const data = await response.json();
+      if (!response.ok) throw new Error(data.error || `Unable to summarize ${filename}`);
+      results.push({ filename, summary: data.summary, tags: data.tags });
+    }
+    advisorContextSummary.innerHTML = results
+      .map(
+        (item) => `
+          <div class="summary-block">
+            <h4 class="eyebrow">${item.filename}</h4>
+            <p class="text-muted">${item.summary || 'No summary'}</p>
+            <div class="tag-cloud">${(item.tags || []).map((tag) => `<span class="badge">${tag}</span>`).join('')}</div>
+          </div>
+        `,
+      )
+      .join('');
+  } catch (error) {
+    advisorContextSummary.innerHTML = `<div class="inline-alert error">${error.message}</div>`;
+  }
+}
+
+chatForm?.addEventListener('submit', (event) => {
+  event.preventDefault();
+  const message = chatPrompt.value.trim();
+  if (!message) return;
+  sendMessage(message);
+});
+
+loadContextBtn?.addEventListener('click', (event) => {
+  event.preventDefault();
+  loadSummaries();
+});
+
+promptChips.forEach((chip) => {
+  chip.addEventListener('click', () => {
+    const prompt = chip.getAttribute('data-prompt');
+    if (!prompt) return;
+    chatPrompt.value = prompt;
+    chatPrompt.focus();
+  });
+});
diff --git a/app/static/js/comparison.js b/app/static/js/comparison.js
new file mode 100644
index 0000000..4102589
--- /dev/null
+++ b/app/static/js/comparison.js
@@ -0,0 +1,96 @@
+const leftDoc = document.getElementById('leftDoc');
+const rightDoc = document.getElementById('rightDoc');
+const compareBtn = document.getElementById('compareBtn');
+const comparisonOutput = document.getElementById('comparisonOutput');
+const comparisonDial = document.getElementById('comparisonDial');
+const comparisonScore = document.getElementById('comparisonScore');
+const comparisonSynopsis = document.getElementById('comparisonSynopsis');
+
+function updateDial(score) {
+  const numeric = Number.parseInt(score, 10) || 0;
+  const bounded = Math.max(0, Math.min(numeric, 100));
+  comparisonDial?.style.setProperty('--score', bounded);
+  comparisonDial?.setAttribute('data-score', bounded);
+  if (comparisonScore) {
+    comparisonScore.textContent = Number.isFinite(numeric) ? bounded : '—';
+  }
+}
+
+function renderComparison(data) {
+  comparisonSynopsis.textContent = data.synopsis || 'No synopsis available.';
+  updateDial(data.comparison_score ?? 0);
+  comparisonOutput.innerHTML = `
+    <div class="comparison-grid">
+      <section class="difference-group">
+        <h4>Alignment</h4>
+        ${renderList(data.alignment, 'theme', 'assessment', false)}
+      </section>
+      <section class="difference-group">
+        <h4>Divergence</h4>
+        ${renderList(data.divergence, 'issue', 'detail', true)}
+      </section>
+      <section class="difference-group">
+        <h4>Recommendations</h4>
+        ${renderRecommendations(data.recommendations)}
+      </section>
+    </div>
+  `;
+}
+
+function renderList(items = [], primaryKey, secondaryKey, includeSeverity = false) {
+  if (!items.length) {
+    return '<p class="text-muted">No items available.</p>';
+  }
+  return (
+    '<ul class="list-stack">' +
+    items
+      .map((item) => {
+        const severity = includeSeverity && item.severity ? `<span class="badge">${item.severity}</span>` : '';
+        return `
+          <li class="list-stack__item">
+            <div class="summary-block">
+              <strong>${item[primaryKey] || ''}</strong> ${severity}
+              <p class="text-muted">${item[secondaryKey] || ''}</p>
+            </div>
+          </li>
+        `;
+      })
+      .join('') +
+    '</ul>'
+  );
+}
+
+function renderRecommendations(items = []) {
+  if (!items.length) {
+    return '<p class="text-muted">No recommendations provided.</p>';
+  }
+  return (
+    '<ul class="list-stack">' +
+    items
+      .map((item) => `<li class="list-stack__item">${item}</li>`)
+      .join('') +
+    '</ul>'
+  );
+}
+
+compareBtn?.addEventListener('click', async () => {
+  const left = leftDoc.value;
+  const right = rightDoc.value;
+  if (!left || !right) {
+    comparisonOutput.innerHTML = '<div class="inline-alert warning">Choose two documents to compare.</div>';
+    return;
+  }
+  comparisonOutput.innerHTML = '<div class="loading-spinner" aria-hidden="true"></div>';
+  try {
+    const response = await fetch('/api/compare', {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify({ left, right }),
+    });
+    const data = await response.json();
+    if (!response.ok) throw new Error(data.error || 'Unable to compare documents');
+    renderComparison(data);
+  } catch (error) {
+    comparisonOutput.innerHTML = `<div class="inline-alert error">${error.message}</div>`;
+  }
+});
diff --git a/app/static/js/core.js b/app/static/js/core.js
new file mode 100644
index 0000000..1f9220b
--- /dev/null
+++ b/app/static/js/core.js
@@ -0,0 +1,99 @@
+const commandElements = document.querySelectorAll('[data-command]');
+const drawer = document.getElementById('playbookDrawer');
+const navToggle = document.getElementById('navToggle');
+const shellNav = document.getElementById('shellNav');
+let toastTimeout;
+
+function showNotification(message) {
+  if (!message) return;
+  let toast = document.querySelector('.aurora-toast');
+  if (!toast) {
+    toast = document.createElement('div');
+    toast.className = 'aurora-toast';
+    document.body.appendChild(toast);
+  }
+  toast.textContent = message;
+  toast.classList.add('visible');
+  clearTimeout(toastTimeout);
+  toastTimeout = setTimeout(() => {
+    toast.classList.remove('visible');
+  }, 2800);
+}
+
+function toggleFocusMode() {
+  document.body.classList.toggle('focus-mode');
+  const enabled = document.body.classList.contains('focus-mode');
+  showNotification(enabled ? 'Focus mode enabled' : 'Focus mode disabled');
+}
+
+function syncCatalog() {
+  document.dispatchEvent(new CustomEvent('aurora:refresh-catalog'));
+  showNotification('Catalog sync triggered');
+}
+
+function toggleDrawer(forceOpen = null) {
+  if (!drawer) return;
+  const shouldOpen = forceOpen ?? !drawer.classList.contains('open');
+  drawer.classList.toggle('open', shouldOpen);
+}
+
+commandElements.forEach((element) => {
+  element.addEventListener('click', (event) => {
+    event.preventDefault();
+    const command = event.currentTarget.getAttribute('data-command');
+    switch (command) {
+      case 'focus':
+        toggleFocusMode();
+        break;
+      case 'refresh':
+        syncCatalog();
+        break;
+      case 'docs':
+        toggleDrawer(true);
+        break;
+      case 'close-drawer':
+        toggleDrawer(false);
+        break;
+      default:
+        break;
+    }
+  });
+});
+
+if (drawer) {
+  drawer.addEventListener('click', (event) => {
+    if (event.target === drawer) {
+      toggleDrawer(false);
+    }
+  });
+}
+
+navToggle?.addEventListener('click', () => {
+  const expanded = navToggle.getAttribute('aria-expanded') === 'true';
+  const shouldOpen = !expanded;
+  navToggle.setAttribute('aria-expanded', String(shouldOpen));
+  shellNav?.classList.toggle('open', shouldOpen);
+});
+
+if (shellNav) {
+  shellNav.querySelectorAll('a').forEach((link) => {
+    link.addEventListener('click', () => {
+      navToggle?.setAttribute('aria-expanded', 'false');
+      shellNav.classList.remove('open');
+    });
+  });
+}
+
+document.addEventListener('keydown', (event) => {
+  if (event.ctrlKey && !event.shiftKey && event.key.toLowerCase() === 'b') {
+    event.preventDefault();
+    toggleFocusMode();
+  }
+  if (event.ctrlKey && event.altKey && event.key.toLowerCase() === 'r') {
+    event.preventDefault();
+    syncCatalog();
+  }
+  if (event.key === 'Escape' && drawer?.classList.contains('open')) {
+    toggleDrawer(false);
+  }
+});
diff --git a/app/static/js/data-management.js b/app/static/js/data-management.js
new file mode 100644
index 0000000..5503f80
--- /dev/null
+++ b/app/static/js/data-management.js
@@ -0,0 +1,131 @@
+const catalogList = document.getElementById('catalogList');
+const refreshCatalogBtn = document.getElementById('refreshCatalog');
+const summarySelect = document.getElementById('summarySelect');
+const summaryOutput = document.getElementById('summaryOutput');
+const qualityDial = document.getElementById('documentScore');
+
+async function refreshCatalog() {
+  if (!catalogList) return;
+  refreshCatalogBtn?.classList.add('disabled');
+  try {
+    const response = await fetch('/api/catalog');
+    if (!response.ok) throw new Error('Failed to load catalog');
+    const data = await response.json();
+    catalogList.innerHTML = buildCatalogGrid(data.documents || []);
+    wirePreviewButtons();
+    if (summarySelect) {
+      const current = summarySelect.value;
+      summarySelect.innerHTML = '<option value="">Choose…</option>' +
+        (data.documents || [])
+          .map((doc) => `<option value="${doc.name}">${doc.name}</option>`)
+          .join('');
+      if (current) {
+        summarySelect.value = current;
+      }
+    }
+  } catch (error) {
+    catalogList.innerHTML = `<div class="inline-alert error">${error.message}</div>`;
+  } finally {
+    refreshCatalogBtn?.classList.remove('disabled');
+  }
+}
+
+function buildCatalogGrid(documents) {
+  if (!documents.length) {
+    return '<p class="text-muted">No documents yet. Upload a file to populate the catalog.</p>';
+  }
+  return documents
+    .map(
+      (doc) => `
+        <article class="catalog-card" data-filename="${doc.name}">
+          <header>
+            <span class="badge">${doc.type || 'DOC'}</span>
+            <span>${doc.updated}</span>
+          </header>
+          <h3>${doc.name}</h3>
+          <p class="text-muted">${doc.size}</p>
+          <div class="catalog-actions">
+            <button type="button" class="btn btn--outline btn--small" data-preview="${doc.name}">Profile</button>
+          </div>
+        </article>
+      `,
+    )
+    .join('');
+}
+
+function updateQualityDial(score) {
+  if (!qualityDial) return;
+  const value = qualityDial.querySelector('.value');
+  const numericScore = typeof score === 'number' ? score : parseInt(score, 10) || 0;
+  qualityDial.dataset.score = numericScore;
+  qualityDial.style.setProperty('--score', Math.max(0, Math.min(numericScore, 100)));
+  if (value) {
+    value.textContent = Number.isFinite(numericScore) ? `${numericScore}` : '—';
+  }
+}
+
+function renderSummary(payload) {
+  summaryOutput.innerHTML = `
+    <div class="summary-grid">
+      <div class="summary-block">
+        <h3 class="display-md">${payload.summary || 'Summary unavailable'}</h3>
+        <p class="text-muted">Grounded in uploaded context with AI-generated highlights.</p>
+      </div>
+      <div class="summary-block">
+        <h4 class="eyebrow">Tags</h4>
+        <div class="tag-cloud">${(payload.tags || []).map((tag) => `<span class="badge">${tag}</span>`).join('') || '<span class="text-muted">No tags provided</span>'}</div>
+      </div>
+      <div class="summary-block">
+        <h4 class="eyebrow">Opportunities</h4>
+        <ul class="list-stack">${(payload.opportunities || []).map((item) => `<li class="list-stack__item">${item}</li>`).join('') || '<li class="list-stack__item">None detected</li>'}</ul>
+      </div>
+    </div>
+  `;
+  updateQualityDial(payload.data_quality ?? 0);
+}
+
+async function summarizeDocument(filename) {
+  if (!filename) {
+    summaryOutput.innerHTML = '<p class="text-muted">Summaries will appear here once generated.</p>';
+    updateQualityDial(0);
+    return;
+  }
+  summaryOutput.innerHTML = '<div class="loading-spinner" aria-hidden="true"></div>';
+  try {
+    const response = await fetch('/api/data/summarize', {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify({ filename }),
+    });
+    const payload = await response.json();
+    if (!response.ok) throw new Error(payload.error || 'Unable to summarize document');
+    renderSummary(payload);
+  } catch (error) {
+    summaryOutput.innerHTML = `<div class="inline-alert error">${error.message}</div>`;
+    updateQualityDial(0);
+  }
+}
+
+function wirePreviewButtons() {
+  if (!catalogList) return;
+  catalogList.querySelectorAll('[data-preview]').forEach((button) => {
+    button.addEventListener('click', (event) => {
+      const name = event.currentTarget.getAttribute('data-preview');
+      if (summarySelect) {
+        summarySelect.value = name;
+      }
+      summarizeDocument(name);
+    });
+  });
+}
+
+refreshCatalogBtn?.addEventListener('click', refreshCatalog);
+summarySelect?.addEventListener('change', (event) => summarizeDocument(event.target.value));
+
+document.addEventListener('aurora:refresh-catalog', refreshCatalog);
+
+wirePreviewButtons();
+
+if (summarySelect?.value) {
+  summarizeDocument(summarySelect.value);
+}
diff --git a/app/static/js/prediction.js b/app/static/js/prediction.js
new file mode 100644
index 0000000..e77cfd4
--- /dev/null
+++ b/app/static/js/prediction.js
@@ -0,0 +1,144 @@
+const predictionForm = document.getElementById('predictionForm');
+const predictionSummary = document.getElementById('predictionSummary');
+const insightCards = document.getElementById('insightCards');
+const initiativeList = document.getElementById('initiativeList');
+const confidenceBadge = document.getElementById('confidenceBadge');
+const outlookDial = document.getElementById('outlookDial');
+const outlookValue = document.getElementById('outlookValue');
+const runwayIndicator = document.getElementById('runwayIndicator');
+const metricHealth = document.getElementById('metricHealth');
+const impactGrid = document.getElementById('impactGrid');
+
+function extractFormData(form) {
+  const formData = new FormData(form);
+  const values = Object.fromEntries(formData.entries());
+  values.documents = formData.getAll('documents');
+  return values;
+}
+
+function renderImpactGrid(dataset) {
+  if (!impactGrid) return;
+  const labels = dataset?.chart?.labels || [];
+  const series = dataset?.chart?.series || [];
+  if (!labels.length || !series.length) {
+    impactGrid.innerHTML = '<p class="text-muted">No impact metrics provided.</p>';
+    return;
+  }
+  impactGrid.innerHTML = labels
+    .map((label, index) => {
+      const value = Number.parseFloat(series[index]) || 0;
+      const bounded = Math.max(0, Math.min(value, 100));
+      return `
+        <div class="coverage-bar">
+          <span class="text-muted">${label}</span>
+          <div class="bar"><span style="width: ${bounded}%"></span></div>
+          <small class="text-muted">${value.toFixed(1)}</small>
+        </div>
+      `;
+    })
+    .join('');
+}
+
+function updateOutlook(score) {
+  if (!outlookDial) return;
+  const numeric = Number.parseInt(score, 10) || 0;
+  const bounded = Math.max(0, Math.min(numeric, 100));
+  outlookDial.dataset.score = bounded;
+  outlookDial.style.setProperty('--score', bounded);
+  if (outlookValue) {
+    outlookValue.textContent = Number.isFinite(bounded) ? bounded : '—';
+  }
+  if (!runwayIndicator || !metricHealth) return;
+  if (bounded >= 70) {
+    runwayIndicator.textContent = 'Accelerate';
+    metricHealth.textContent = 'Momentum suggests aggressive plays.';
+  } else if (bounded >= 40) {
+    runwayIndicator.textContent = 'Stabilize';
+    metricHealth.textContent = 'Blend growth with retention safeguards.';
+  } else {
+    runwayIndicator.textContent = 'Recalibrate';
+    metricHealth.textContent = 'Revisit fundamentals and cost structure.';
+  }
+}
+
+function renderInsightCards(cards = []) {
+  if (!cards.length) {
+    insightCards.innerHTML = '<p class="text-muted">No insights yet.</p>';
+    return;
+  }
+  insightCards.innerHTML = cards
+    .map(
+      (card) => `
+        <div class="insight-pill">
+          <h3>${card.title || 'Insight'}</h3>
+          <p class="text-muted">${card.detail || ''}</p>
+        </div>
+      `,
+    )
+    .join('');
+}
+
+function renderInitiatives(items = []) {
+  if (!items.length) {
+    initiativeList.innerHTML = '<li class="list-stack__item">No initiatives supplied.</li>';
+    return;
+  }
+  initiativeList.innerHTML = items.map((item) => `<li class="list-stack__item">${item}</li>`).join('');
+}
+
+function applyTemplate(button) {
+  const template = button.getAttribute('data-template');
+  if (!template) return;
+  try {
+    const values = JSON.parse(template);
+    Object.entries(values).forEach(([key, value]) => {
+      const field = predictionForm?.elements.namedItem(key);
+      if (!field) return;
+      if (field instanceof HTMLSelectElement || field instanceof HTMLInputElement || field instanceof HTMLTextAreaElement) {
+        field.value = value;
+      }
+    });
+  } catch (error) {
+    console.error('Unable to apply template', error);
+  }
+}
+
+predictionForm?.addEventListener('submit', async (event) => {
+  event.preventDefault();
+  predictionSummary.innerHTML = '<div class="loading-spinner" aria-hidden="true"></div>';
+  insightCards.innerHTML = '';
+  initiativeList.innerHTML = '';
+  confidenceBadge.textContent = '—';
+  runwayIndicator.textContent = '—';
+  metricHealth.textContent = 'Awaiting forecast';
+  updateOutlook(0);
+  try {
+    const payload = extractFormData(predictionForm);
+    const response = await fetch('/api/predict', {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify(payload),
+    });
+    const data = await response.json();
+    if (!response.ok) throw new Error(data.error || 'Unable to generate forecast');
+    predictionSummary.textContent = data.summary || 'No summary provided.';
+    const confidence = data.confidence ?? '—';
+    const numericConfidence = Number.parseFloat(confidence);
+    confidenceBadge.textContent = Number.isFinite(numericConfidence)
+      ? `${numericConfidence} / 100`
+      : confidence;
+    updateOutlook(data.growth_outlook ?? 0);
+    renderInsightCards(data.insight_cards || []);
+    renderInitiatives(data.key_initiatives || []);
+    renderImpactGrid(data);
+  } catch (error) {
+    predictionSummary.innerHTML = `<div class="inline-alert error">${error.message}</div>`;
+    renderImpactGrid({ chart: { labels: [], series: [] } });
+  }
+});
+
+document.querySelectorAll('.template-chip').forEach((button) => {
+  button.addEventListener('click', () => applyTemplate(button));
+});
+
+renderImpactGrid();
diff --git a/app/static/js/risk.js b/app/static/js/risk.js
new file mode 100644
index 0000000..9cc9296
--- /dev/null
+++ b/app/static/js/risk.js
@@ -0,0 +1,118 @@
+const riskForm = document.getElementById('riskForm');
+const riskSummary = document.getElementById('riskSummary');
+const riskMatrix = document.getElementById('riskMatrix');
+const warningList = document.getElementById('warningList');
+const riskScoreValue = document.getElementById('riskScore');
+const riskDial = document.getElementById('riskDial');
+const riskPriority = document.getElementById('riskPriority');
+const warningCount = document.getElementById('warningCount');
+const exposureGrid = document.getElementById('exposureGrid');
+
+function getRiskFormData(form) {
+  const formData = new FormData(form);
+  const payload = Object.fromEntries(formData.entries());
+  payload.documents = formData.getAll('riskDocuments');
+  return payload;
+}
+
+function renderExposureGrid(heatmap = {}) {
+  if (!exposureGrid) return;
+  const labels = heatmap.labels || [];
+  const series = heatmap.series || [];
+  if (!labels.length || !series.length) {
+    exposureGrid.innerHTML = '<p class="text-muted">No exposure distribution provided.</p>';
+    return;
+  }
+  exposureGrid.innerHTML = labels
+    .map((label, index) => {
+      const value = Number.parseFloat(series[index]) || 0;
+      const bounded = Math.max(0, Math.min(value, 100));
+      return `
+        <div class="coverage-bar">
+          <span class="text-muted">${label}</span>
+          <div class="bar"><span style="width: ${bounded}%"></span></div>
+          <small class="text-muted">${value.toFixed(1)}</small>
+        </div>
+      `;
+    })
+    .join('');
+}
+
+function renderMatrix(items = []) {
+  if (!items.length) {
+    riskMatrix.innerHTML = '<p class="text-muted">No risks surfaced yet.</p>';
+    return;
+  }
+  riskMatrix.innerHTML = items
+    .map((item) => {
+      const mitigation = item.mitigation ? `<span class="badge">${item.mitigation}</span>` : '';
+      return `
+        <div class="insight-pill">
+          <h4>${item.name || 'Risk'}</h4>
+          <p class="text-muted">Likelihood <strong>${item.likelihood ?? '—'}</strong> · Impact <strong>${item.impact ?? '—'}</strong></p>
+          <p class="text-muted">${item.detail || ''}</p>
+          ${mitigation}
+        </div>
+      `;
+    })
+    .join('');
+}
+
+function renderWarnings(items = []) {
+  if (!items.length) {
+    warningList.innerHTML = '<li class="list-stack__item">No immediate warnings detected.</li>';
+    warningCount.textContent = '0';
+    return;
+  }
+  warningList.innerHTML = items.map((item) => `<li class="list-stack__item">${item}</li>`).join('');
+  warningCount.textContent = items.length;
+}
+
+function updateRiskDial(score) {
+  const numeric = Number.parseInt(score, 10) || 0;
+  const bounded = Math.max(0, Math.min(numeric, 100));
+  riskDial?.style.setProperty('--score', bounded);
+  riskDial?.setAttribute('data-score', bounded);
+  if (riskScoreValue) {
+    riskScoreValue.textContent = Number.isFinite(bounded) ? bounded : '—';
+  }
+}
+
+function extractPriority(items = []) {
+  if (!items.length) {
+    riskPriority.textContent = '—';
+    return;
+  }
+  const prioritized = [...items].sort((a, b) => (Number(b.impact) || 0) - (Number(a.impact) || 0));
+  riskPriority.textContent = prioritized[0]?.name || '—';
+}
+
+riskForm?.addEventListener('submit', async (event) => {
+  event.preventDefault();
+  riskSummary.innerHTML = '<div class="loading-spinner" aria-hidden="true"></div>';
+  riskMatrix.innerHTML = '';
+  warningList.innerHTML = '';
+  warningCount.textContent = '—';
+  updateRiskDial(0);
+  try {
+    const payload = getRiskFormData(riskForm);
+    const response = await fetch('/api/risk', {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify(payload),
+    });
+    const data = await response.json();
+    if (!response.ok) throw new Error(data.error || 'Unable to complete risk analysis');
+    riskSummary.textContent = data.risk_summary || 'No summary provided.';
+    updateRiskDial(data.overall_score ?? 0);
+    renderExposureGrid(data.heatmap || {});
+    renderMatrix(data.risk_matrix || []);
+    extractPriority(data.risk_matrix || []);
+    renderWarnings(data.early_warnings || []);
+  } catch (error) {
+    riskSummary.innerHTML = `<div class="inline-alert error">${error.message}</div>`;
+    renderExposureGrid({ labels: [], series: [] });
+  }
+});
+
+renderExposureGrid();
diff --git a/app/templates/advisor.html b/app/templates/advisor.html
new file mode 100644
index 0000000..583e319
--- /dev/null
+++ b/app/templates/advisor.html
@@ -0,0 +1,61 @@
+{% extends 'base.html' %}
+{% block title %}Advisor Lounge · Aurora CRM{% endblock %}
+{% block content %}
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h1 class="display-md">Advisor lounge</h1>
+      <p class="text-muted">Chat with the embedded strategist for pitch angles, rollout plans, and follow-up actions. Add document context to ground the guidance.</p>
+    </div>
+    <div>
+      <div class="kpi-card">
+        <span class="kpi-label">Response status</span>
+        <span class="kpi-value" id="chatStatus">Ready</span>
+        <p class="text-muted">Local model orchestrates every reply.</p>
+      </div>
+    </div>
+  </div>
+</section>
+
+<section class="panel panel--soft">
+  <div class="chat-wrapper">
+    <div>
+      <div class="chat-log" id="chatLog"></div>
+      <form id="chatForm" class="chat-console">
+        <label class="form-field">
+          <span class="form-label">Prompt</span>
+          <textarea id="chatPrompt" class="form-control" placeholder="Ask for a launch plan, executive summary, or follow-up actions" required></textarea>
+        </label>
+        <div class="template-tray">
+          <button type="button" class="prompt-chip" data-prompt="Draft a three-point executive summary for the current pipeline outlook.">Executive summary</button>
+          <button type="button" class="prompt-chip" data-prompt="What should we do next to improve retention for our enterprise cohort?">Retention booster</button>
+          <button type="button" class="prompt-chip" data-prompt="List critical risks we need to brief the leadership team on this week.">Leadership brief</button>
+        </div>
+        <button class="btn btn--primary" type="submit">Send</button>
+      </form>
+    </div>
+    <div class="ai-panel">
+      <h2 class="display-md">Context vault</h2>
+      <label class="form-field">
+        <span class="form-label">Conversation context</span>
+        <textarea class="form-control" id="advisorContext" rows="4" placeholder="Paste CRM snippets or summary notes to anchor the advisor."></textarea>
+      </label>
+      <label class="form-field">
+        <span class="form-label">Attach catalog documents</span>
+        <select class="form-select" id="advisorDocuments" multiple size="8">
+          {% for doc in catalog %}
+            <option value="{{ doc.name }}">{{ doc.name }}</option>
+          {% endfor %}
+        </select>
+      </label>
+      <button class="btn btn--outline" id="loadContext">Load quick summaries</button>
+      <div class="ai-panel" id="advisorContextSummary">
+        <p class="text-muted">Summaries will appear here after loading documents.</p>
+      </div>
+    </div>
+  </div>
+</section>
+{% endblock %}
+{% block scripts %}
+  <script src="{{ url_for('static', filename='js/advisor.js') }}"></script>
+{% endblock %}
diff --git a/app/templates/base.html b/app/templates/base.html
new file mode 100644
index 0000000..ebea236
--- /dev/null
+++ b/app/templates/base.html
@@ -0,0 +1,116 @@
+<!doctype html>
+<html lang="en">
+  <head>
+    <meta charset="utf-8" />
+    <meta name="viewport" content="width=device-width, initial-scale=1" />
+    <title>{% block title %}Aurora CRM Intelligence{% endblock %}</title>
+    <link rel="stylesheet" href="{{ url_for('static', filename='css/styles.css') }}" />
+    {% block head %}{% endblock %}
+  </head>
+  <body>
+    <div class="background-veil"></div>
+    <header class="shell-header">
+      <div class="container shell-header__inner">
+        <a class="brand-mark" href="{{ url_for('core.overview') }}">
+          <span class="brand-icon">A</span>
+          Aurora CRM
+        </a>
+        <div class="shell-header__actions">
+          <button class="command-btn" data-command="focus">Focus</button>
+          <button class="command-btn" data-command="refresh">Sync</button>
+          <button class="command-btn" data-command="docs">Playbook</button>
+          <button class="nav-toggle" id="navToggle" aria-expanded="false" aria-controls="shellNav">☰</button>
+        </div>
+      </div>
+      <nav class="container shell-nav" id="shellNav">
+        <a class="shell-nav__link {% if request.endpoint == 'core.overview' %}is-active{% endif %}" href="{{ url_for('core.overview') }}">Overview</a>
+        <a class="shell-nav__link {% if request.endpoint == 'core.data_management' %}is-active{% endif %}" href="{{ url_for('core.data_management') }}">Data</a>
+        <a class="shell-nav__link {% if request.endpoint == 'core.predictive_strategy' %}is-active{% endif %}" href="{{ url_for('core.predictive_strategy') }}">Prediction</a>
+        <a class="shell-nav__link {% if request.endpoint == 'core.risk_command' %}is-active{% endif %}" href="{{ url_for('core.risk_command') }}">Risk</a>
+        <a class="shell-nav__link {% if request.endpoint == 'core.ai_advisor' %}is-active{% endif %}" href="{{ url_for('core.ai_advisor') }}">AI Advisor</a>
+        <a class="shell-nav__link {% if request.endpoint == 'core.comparison_lab' %}is-active{% endif %}" href="{{ url_for('core.comparison_lab') }}">Comparison</a>
+      </nav>
+    </header>
+
+    <section class="status-strip">
+      <div class="container status-strip__inner">
+        <div class="status-grid">
+          <article class="status-card">
+            <span class="status-card__label">Neural core</span>
+            <div class="status-card__value {{ 'online' if api_ready else 'offline' }}">
+              <span class="dot"></span>
+              {{ 'Local model ready' if api_ready else 'Model offline' }}
+            </div>
+            <p class="status-card__hint">
+              {% if api_ready %}
+                Running inference from your GPU-hosted model.
+              {% else %}
+                Launch Ollama and configure <code>LLM_MODEL</code> in your <code>.env</code> file.
+              {% endif %}
+            </p>
+          </article>
+          <article class="status-card">
+            <span class="status-card__label">Catalog</span>
+            <div class="status-card__value">
+              {{ catalog_summary.count if catalog_summary else 0 }} files
+            </div>
+            <p class="status-card__hint">{{ catalog_summary.total_size if catalog_summary else '0 B' }} indexed footprint</p>
+          </article>
+          <article class="status-card">
+            <span class="status-card__label">Median artifact</span>
+            <div class="status-card__value">
+              {{ catalog_summary.average_size if catalog_summary else '0 B' }}
+            </div>
+            <p class="status-card__hint">Coverage across {{ catalog_summary.coverage|length if catalog_summary else 0 }} sources</p>
+          </article>
+        </div>
+        <div class="command-strip">
+          <a class="command-btn" href="{{ url_for('core.data_management') }}">Upload data</a>
+          <button class="command-btn" data-command="focus">Toggle focus</button>
+          <button class="command-btn" data-command="refresh">Refresh catalog</button>
+          <button class="command-btn" data-command="docs">Open playbook</button>
+        </div>
+      </div>
+    </section>
+
+    <div class="playbook-drawer" id="playbookDrawer">
+      <div class="playbook-panel">
+        <header>
+          <div>
+            <span class="hero-eyebrow"><span class="dot"></span>Playbook</span>
+            <h2 class="display-md">Power-user shortcuts</h2>
+          </div>
+          <button class="icon-btn" data-command="close-drawer" aria-label="Close">×</button>
+        </header>
+        <div>
+          <h3 class="kpi-label">Instant actions</h3>
+          <ul class="timeline">
+            <li class="timeline-item">Upload CRM context from the Data screen to expand the intelligence corpus.</li>
+            <li class="timeline-item">Run Predictive Strategy after each pipeline update to recalibrate growth tracks.</li>
+            <li class="timeline-item">Use Comparison Lab before executive reviews to surface deltas and alignment gaps.</li>
+          </ul>
+        </div>
+        <div>
+          <h3 class="kpi-label">Keyboard tips</h3>
+          <p class="text-muted">Focus mode toggles with <kbd>Ctrl</kbd> + <kbd>B</kbd>. Refresh catalog with <kbd>Ctrl</kbd> + <kbd>Alt</kbd> + <kbd>R</kbd>.</p>
+        </div>
+      </div>
+    </div>
+
+    <main class="container page-shell">
+      {% with messages = get_flashed_messages(with_categories=true) %}
+        {% if messages %}
+          <div class="flash-stack">
+            {% for category, message in messages %}
+              <div class="flash-message {{ category }}">{{ message }}</div>
+            {% endfor %}
+          </div>
+        {% endif %}
+      {% endwith %}
+      {% block content %}{% endblock %}
+    </main>
+
+    <script src="{{ url_for('static', filename='js/core.js') }}"></script>
+    {% block scripts %}{% endblock %}
+  </body>
+</html>
diff --git a/app/templates/comparison.html b/app/templates/comparison.html
new file mode 100644
index 0000000..ad6f6b6
--- /dev/null
+++ b/app/templates/comparison.html
@@ -0,0 +1,65 @@
+{% extends 'base.html' %}
+{% block title %}Comparison Lab · Aurora CRM{% endblock %}
+{% block content %}
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h1 class="display-md">Deep comparison lab</h1>
+      <p class="text-muted">Select any two catalogued documents to pinpoint alignment, divergence, and recommended remediation – all computed locally.</p>
+    </div>
+    <div>
+      <div class="kpi-card">
+        <span class="kpi-label">Local inference</span>
+        <span class="kpi-value">{{ 'Active' if api_ready else 'Offline' }}</span>
+        <p class="text-muted">GPU-powered differential analysis on your machine.</p>
+      </div>
+    </div>
+  </div>
+</section>
+
+<section class="panel panel--soft">
+  <div class="form-grid" style="grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));">
+    <label class="form-field">
+      <span class="form-label">Document A</span>
+      <select class="form-select" id="leftDoc">
+        <option value="">Choose…</option>
+        {% for doc in catalog %}
+          <option value="{{ doc.name }}">{{ doc.name }}</option>
+        {% endfor %}
+      </select>
+    </label>
+    <label class="form-field">
+      <span class="form-label">Document B</span>
+      <select class="form-select" id="rightDoc">
+        <option value="">Choose…</option>
+        {% for doc in catalog %}
+          <option value="{{ doc.name }}">{{ doc.name }}</option>
+        {% endfor %}
+      </select>
+    </label>
+    <div class="form-field" style="align-self: end;">
+      <button class="btn btn--primary btn--full" id="compareBtn">Compare</button>
+    </div>
+  </div>
+  <div class="metric-board" style="margin-top: 1.5rem;">
+    <div class="metric-card">
+      <span class="label">Alignment index</span>
+      <div class="metric-dial" id="comparisonDial" data-score="0">
+        <div class="value" id="comparisonScore">0</div>
+      </div>
+      <p class="text-muted">Higher is better alignment</p>
+    </div>
+    <div class="metric-card">
+      <span class="label">Synopsis</span>
+      <div class="value" id="comparisonSynopsis">Awaiting analysis</div>
+      <p class="text-muted">High-level summary</p>
+    </div>
+  </div>
+  <div class="ai-panel" id="comparisonOutput" style="margin-top: 1.5rem;">
+    <p class="text-muted">Select two files to run a deep comparison.</p>
+  </div>
+</section>
+{% endblock %}
+{% block scripts %}
+  <script src="{{ url_for('static', filename='js/comparison.js') }}"></script>
+{% endblock %}
diff --git a/app/templates/data_management.html b/app/templates/data_management.html
new file mode 100644
index 0000000..d8cf95b
--- /dev/null
+++ b/app/templates/data_management.html
@@ -0,0 +1,99 @@
+{% extends 'base.html' %}
+{% block title %}Data Steward · Aurora CRM{% endblock %}
+{% block content %}
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h1 class="display-md">Curate your CRM intelligence</h1>
+      <p class="text-muted">
+        Upload structured notes, CSV exports, enablement decks, or customer research. Everything stays local while the GPU-backed
+        model builds richer predictions, risks, and advisor summaries.
+      </p>
+      <div class="tag-cloud">
+        <span class="badge">{{ catalog_summary.count if catalog_summary else 0 }} files</span>
+        <span class="badge">{{ catalog_summary.total_size if catalog_summary else '0 B' }} footprint</span>
+        <span class="badge">Median {{ catalog_summary.average_size if catalog_summary else '0 B' }}</span>
+      </div>
+    </div>
+    <div>
+      <div class="kpi-card">
+        <span class="kpi-label">Neural status</span>
+        <span class="kpi-value">{{ 'Ready' if api_ready else 'Offline' }}</span>
+        <p class="text-muted">
+          {% if api_ready %}
+            Model streaming from your 4090 – no external API calls required.
+          {% else %}
+            Start <code>ollama</code> and set <code>LLM_MODEL</code> to a pulled model such as <code>llama3.2:latest</code>.
+          {% endif %}
+        </p>
+      </div>
+    </div>
+  </div>
+</section>
+
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h2 class="display-md">Upload document</h2>
+      <p class="text-muted">Supported formats: TXT, MD, CSV, JSON. Files remain on disk inside the <code>uploads</code> folder.</p>
+      <form action="{{ url_for('core.upload_file') }}" method="post" enctype="multipart/form-data" class="form-grid" id="uploadForm">
+        <label class="form-field">
+          <span class="form-label">Choose file</span>
+          <input class="form-control" type="file" id="document" name="document" required />
+        </label>
+        <button class="btn btn--primary" type="submit">Add to catalog</button>
+      </form>
+      <div class="story-card" style="margin-top: 1.5rem;">
+        <h3>Tips</h3>
+        <ul>
+          <li>Use descriptive filenames so the advisor can cite sources.</li>
+          <li>Upload paired documents to unlock deeper comparisons.</li>
+          <li>Trigger a catalog refresh after large batches.</li>
+        </ul>
+      </div>
+    </div>
+    <div>
+      <div class="section-heading">
+        <div>
+          <p class="eyebrow">Catalog</p>
+          <h3 class="display-md">Your local corpus</h3>
+        </div>
+        <button class="btn btn--outline btn--small" id="refreshCatalog">Refresh</button>
+      </div>
+      <div id="catalogList" class="catalog-grid">
+        {% include 'partials/catalog_list.html' %}
+      </div>
+    </div>
+  </div>
+</section>
+
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h2 class="display-md">Document profile</h2>
+      <p class="text-muted">Select a file to generate a synopsis, tags, and opportunity prompts powered by the local model.</p>
+      <label class="form-field">
+        <span class="form-label">Document</span>
+        <select class="form-select" id="summarySelect">
+          <option value="">Choose…</option>
+          {% for doc in catalog %}
+            <option value="{{ doc.name }}">{{ doc.name }}</option>
+          {% endfor %}
+        </select>
+      </label>
+      <div class="quality-dial" id="documentScore" data-score="0">
+        <div class="value">—</div>
+        <p class="text-muted">Quality score</p>
+      </div>
+    </div>
+    <div>
+      <div id="summaryOutput" class="ai-panel document-output">
+        <p class="text-muted">Summaries will appear here once generated.</p>
+      </div>
+    </div>
+  </div>
+</section>
+{% endblock %}
+{% block scripts %}
+  <script src="{{ url_for('static', filename='js/data-management.js') }}"></script>
+{% endblock %}
diff --git a/app/templates/overview.html b/app/templates/overview.html
new file mode 100644
index 0000000..73c7d23
--- /dev/null
+++ b/app/templates/overview.html
@@ -0,0 +1,174 @@
+{% extends 'base.html' %}
+{% block title %}Aurora CRM Intelligence{% endblock %}
+{% block content %}
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <span class="hero-eyebrow"><span class="dot"></span>Aurora command</span>
+      <h1 class="display-xl">Run revenue, risk, and relationships from one neural workspace</h1>
+      <p class="text-muted">
+        Bring pipeline exports, field research, and operator notes together. Aurora synthesizes every upload into prediction arcs,
+        mitigation blueprints, advisor-grade briefings, and deep comparisons – entirely on your RTX-powered workstation.
+      </p>
+      {% if not api_ready %}
+        <div class="inline-alert warning" role="alert">
+          Start <code>ollama</code>, run <code>ollama pull llama3.2</code>, and set <code>LLM_MODEL</code> in your <code>.env</code>
+          to unlock the intelligence modules.
+        </div>
+      {% endif %}
+      <div class="hero-actions">
+        <a class="btn btn--primary" href="{{ url_for('core.predictive_strategy') }}">Launch prediction studio</a>
+        <a class="btn btn--outline" href="{{ url_for('core.data_management') }}">Curate catalog</a>
+        <a class="scroll-link" href="#workflow">Explore capabilities</a>
+      </div>
+      <div class="marquee">
+        <div class="marquee-track">
+          <span class="marquee-item">Catalog depth · <strong>{{ catalog_summary.count if catalog_summary else 0 }}</strong> documents</span>
+          <span class="marquee-item">Aggregate footprint · <strong>{{ catalog_summary.total_size if catalog_summary else '0 B' }}</strong></span>
+          {% for segment in (catalog_summary.coverage or [])[:6] %}
+            <span class="marquee-item">{{ segment.label }} coverage · {{ segment.percent }}%</span>
+          {% endfor %}
+        </div>
+      </div>
+      <div class="spotlight-grid">
+        {% if recent_documents %}
+          {% for doc in recent_documents %}
+            <article class="spotlight-card">
+              <header>
+                <span>{{ doc.type }}</span>
+                <span>{{ doc.updated }}</span>
+              </header>
+              <h3>{{ doc.name }}</h3>
+              <p class="text-dimmed">{{ doc.size }}</p>
+            </article>
+          {% endfor %}
+        {% else %}
+          <div class="spotlight-card">
+            <h3>No uploads yet</h3>
+            <p class="text-dimmed">Add discovery notes, renewal playbooks, or pipeline exports to activate the cockpit.</p>
+          </div>
+        {% endif %}
+      </div>
+    </div>
+    <div>
+      <div class="kpi-grid">
+        <div class="kpi-card primary">
+          <span class="kpi-label">Workflows engaged</span>
+          <span class="kpi-value">5</span>
+          <p class="text-muted">Prediction studio · Risk command · Advisor lounge · Comparison lab · Data steward</p>
+        </div>
+        <div class="kpi-card">
+          <span class="kpi-label">Median artifact</span>
+          <span class="kpi-value">{{ catalog_summary.average_size if catalog_summary else '0 B' }}</span>
+          <p class="text-muted">Sized for rapid GPU inference and comparison.</p>
+        </div>
+        <div class="kpi-card">
+          <span class="kpi-label">Advisor responsiveness</span>
+          <span class="kpi-value">~6s</span>
+          <p class="text-muted">Round-trip response target for on-box chat.</p>
+        </div>
+      </div>
+    </div>
+  </div>
+</section>
+
+<section class="panel panel--soft" id="workflow">
+  <div class="section-heading">
+    <div>
+      <p class="eyebrow">Workflow coverage</p>
+      <h2 class="display-md">Operate every intelligence lane without leaving the cockpit</h2>
+    </div>
+    <a class="btn btn--outline btn--small" href="{{ url_for('core.data_management') }}">Open data steward</a>
+  </div>
+  <div class="hero-grid">
+    <div class="story-card">
+      <h3>Intelligence loop</h3>
+      <p class="text-muted">Feed documents, evolve strategy, evaluate risk, and brief stakeholders in one place.</p>
+      <ul>
+        <li>Ingest, tag, and classify uploads for instant retrieval.</li>
+        <li>Project growth arcs with sensitivity sweeps and initiative scoring.</li>
+        <li>Surface exposure, early warnings, and mitigation owners.</li>
+        <li>Generate advisor briefs and alignment deltas on demand.</li>
+      </ul>
+    </div>
+    <div class="story-grid">
+      <div class="story-card">
+        <h3>Decision runway</h3>
+        <p class="text-muted">Sync weekly execution and quarterly planning off the same digital twin.</p>
+        <ul>
+          <li>Scenario diffs spotlight revenue deltas in seconds.</li>
+          <li>Risk sweeps connect mitigation owners with impact timing.</li>
+        </ul>
+      </div>
+      <div class="story-card">
+        <h3>Relationship intelligence</h3>
+        <p class="text-muted">Ground every customer conversation with comparative context and advisor prompts.</p>
+        <ul>
+          <li>Comparison lab reveals overlaps, conflicts, and contradictions.</li>
+          <li>Advisor lounge turns feedback into prioritized plays.</li>
+        </ul>
+      </div>
+      <div class="story-card highlight">
+        <h3>Catalog health</h3>
+        {% if catalog_summary.coverage %}
+          <p class="text-muted">Coverage spans {{ catalog_summary.coverage|length }} archetypes totalling {{ catalog_summary.total_size }}.</p>
+          <div class="coverage-bars">
+            {% for segment in catalog_summary.coverage %}
+              <div class="coverage-bar">
+                <span class="text-muted">{{ segment.label }}</span>
+                <div class="bar"><span style="width: {{ segment.percent }}%"></span></div>
+                <small class="text-muted">{{ segment.value }} files</small>
+              </div>
+            {% endfor %}
+          </div>
+        {% else %}
+          <p class="text-muted">Upload files to unlock coverage analytics and metadata insights.</p>
+        {% endif %}
+      </div>
+    </div>
+  </div>
+</section>
+
+<section class="feature-grid">
+  <article class="feature-card">
+    <div class="feature-icon">📊</div>
+    <h3>Predictive momentum</h3>
+    <p class="text-muted">Model multi-horizon forecasts, detect inflection points, and surface initiatives that move the needle.</p>
+    <div class="tag-cloud">
+      <span class="badge">Scenario scoring</span>
+      <span class="badge">Sensitivity sweeps</span>
+    </div>
+    <a class="btn btn--small btn--outline" href="{{ url_for('core.predictive_strategy') }}">Scenario console</a>
+  </article>
+  <article class="feature-card">
+    <div class="feature-icon">🛡️</div>
+    <h3>Risk command</h3>
+    <p class="text-muted">Translate telemetry into mitigation playbooks with automated triage, scoring, and early warning systems.</p>
+    <div class="tag-cloud">
+      <span class="badge">Exposure radar</span>
+      <span class="badge">Mitigation routing</span>
+    </div>
+    <a class="btn btn--small btn--outline" href="{{ url_for('core.risk_command') }}">Risk dashboard</a>
+  </article>
+  <article class="feature-card">
+    <div class="feature-icon">🤝</div>
+    <h3>Advisor insights</h3>
+    <p class="text-muted">Prompt the embedded strategist for pitch angles, rollout plans, and board-ready narratives.</p>
+    <div class="tag-cloud">
+      <span class="badge">Context aware chat</span>
+      <span class="badge">Action playbooks</span>
+    </div>
+    <a class="btn btn--small btn--outline" href="{{ url_for('core.ai_advisor') }}">Advisor lounge</a>
+  </article>
+  <article class="feature-card">
+    <div class="feature-icon">🧬</div>
+    <h3>Deep comparison lab</h3>
+    <p class="text-muted">Upload two files to receive redline-style differentials, SWOT signals, and alignment recommendations.</p>
+    <div class="tag-cloud">
+      <span class="badge">Delta explorer</span>
+      <span class="badge">SWOT synthesis</span>
+    </div>
+    <a class="btn btn--small btn--outline" href="{{ url_for('core.comparison_lab') }}">Open lab</a>
+  </article>
+</section>
+{% endblock %}
diff --git a/app/templates/partials/catalog_list.html b/app/templates/partials/catalog_list.html
new file mode 100644
index 0000000..c429cfa
--- /dev/null
+++ b/app/templates/partials/catalog_list.html
@@ -0,0 +1,17 @@
+{% if catalog %}
+  {% for doc in catalog %}
+    <article class="catalog-card" data-filename="{{ doc.name }}">
+      <header>
+        <span class="badge">{{ doc.type }}</span>
+        <span>{{ doc.updated }}</span>
+      </header>
+      <h3>{{ doc.name }}</h3>
+      <p class="text-muted">{{ doc.size }}</p>
+      <div class="catalog-actions">
+        <button type="button" class="btn btn--outline btn--small" data-preview="{{ doc.name }}">Profile</button>
+      </div>
+    </article>
+  {% endfor %}
+{% else %}
+  <p class="text-muted">No documents yet. Upload a file to populate the catalog.</p>
+{% endif %}
diff --git a/app/templates/prediction.html b/app/templates/prediction.html
new file mode 100644
index 0000000..c741969
--- /dev/null
+++ b/app/templates/prediction.html
@@ -0,0 +1,108 @@
+{% extends 'base.html' %}
+{% block title %}Predictive Strategy · Aurora CRM{% endblock %}
+{% block content %}
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h1 class="display-md">Predict revenue arcs</h1>
+      <p class="text-muted">Blend qualitative assumptions with catalogued evidence to receive quantified growth outlooks, confidence signals, and initiative guidance.</p>
+      <div class="template-tray">
+        <span class="text-muted">Jump start:</span>
+        <button type="button" class="template-chip" data-template='{"segment":"Enterprise SaaS","currentRevenue":4500000,"targetRevenue":6500000,"timeline":"12-month","strategies":"Upsell premium tier; Expand channel partnerships"}'>Enterprise expansion</button>
+        <button type="button" class="template-chip" data-template='{"segment":"B2B Fintech","currentRevenue":1200000,"targetRevenue":3000000,"timeline":"18-month","strategies":"Launch embedded product; Secure compliance certifications"}'>Fintech acceleration</button>
+      </div>
+    </div>
+    <div>
+      <div class="kpi-card">
+        <span class="kpi-label">Local inference</span>
+        <span class="kpi-value">{{ 'Active' if api_ready else 'Offline' }}</span>
+        <p class="text-muted">Powered by your configured GGUF model for fully local forecasting.</p>
+      </div>
+    </div>
+  </div>
+</section>
+
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h2 class="display-md">Scenario designer</h2>
+      <form id="predictionForm" class="form-grid">
+        <label class="form-field">
+          <span class="form-label">Primary segment</span>
+          <input class="form-control" type="text" id="segment" name="segment" placeholder="SMB SaaS, enterprise retail…" required />
+        </label>
+        <div class="form-grid" style="grid-template-columns: repeat(auto-fit, minmax(160px, 1fr));">
+          <label class="form-field">
+            <span class="form-label">Current ARR (USD)</span>
+            <input class="form-control" type="number" id="currentRevenue" name="currentRevenue" min="0" step="1000" />
+          </label>
+          <label class="form-field">
+            <span class="form-label">Target ARR (USD)</span>
+            <input class="form-control" type="number" id="targetRevenue" name="targetRevenue" min="0" step="1000" />
+          </label>
+        </div>
+        <label class="form-field">
+          <span class="form-label">Timeline</span>
+          <select class="form-select" id="timeline" name="timeline">
+            <option value="6-month">6 month</option>
+            <option value="12-month" selected>12 month</option>
+            <option value="18-month">18 month</option>
+            <option value="24-month">24 month</option>
+          </select>
+        </label>
+        <label class="form-field">
+          <span class="form-label">Key initiatives</span>
+          <textarea class="form-control" id="strategies" name="strategies" rows="4" placeholder="List notable plays, investments, or GTM adjustments"></textarea>
+        </label>
+        <label class="form-field">
+          <span class="form-label">Reference documents</span>
+          <select class="form-select" id="documents" name="documents" multiple size="6">
+            {% for doc in catalog %}
+              <option value="{{ doc.name }}">{{ doc.name }}</option>
+            {% endfor %}
+          </select>
+        </label>
+        <button class="btn btn--primary btn--full" type="submit">Generate outlook</button>
+      </form>
+    </div>
+    <div>
+      <div class="metric-board">
+        <div class="metric-card">
+          <span class="label">Growth outlook</span>
+          <div class="metric-dial" id="outlookDial" data-score="0">
+            <div class="value" id="outlookValue">0</div>
+          </div>
+          <p class="text-muted">Composite momentum score</p>
+        </div>
+        <div class="metric-card">
+          <span class="label">Confidence</span>
+          <div class="value" id="confidenceBadge">—</div>
+          <p class="text-muted">Model-rated scenario certainty</p>
+        </div>
+        <div class="metric-card">
+          <span class="label">Runway focus</span>
+          <div class="value" id="runwayIndicator">—</div>
+          <p class="text-muted" id="metricHealth">Awaiting forecast</p>
+        </div>
+      </div>
+      <div class="ai-panel" id="predictionPanel" style="margin-top: 1.5rem;">
+        <p class="text-muted" id="predictionSummary">Run a scenario to populate insights and priorities.</p>
+        <div class="insight-pillboard" id="insightCards"></div>
+        <div class="hero-grid" style="margin-top: 1rem;">
+          <div class="ai-panel">
+            <h3 class="eyebrow">Initiatives</h3>
+            <ul class="list-stack" id="initiativeList"></ul>
+          </div>
+          <div class="ai-panel">
+            <h3 class="eyebrow">Impact distribution</h3>
+            <div id="impactGrid" class="coverage-bars"></div>
+          </div>
+        </div>
+      </div>
+    </div>
+  </div>
+</section>
+{% endblock %}
+{% block scripts %}
+  <script src="{{ url_for('static', filename='js/prediction.js') }}"></script>
+{% endblock %}
diff --git a/app/templates/risk.html b/app/templates/risk.html
new file mode 100644
index 0000000..4f764db
--- /dev/null
+++ b/app/templates/risk.html
@@ -0,0 +1,82 @@
+{% extends 'base.html' %}
+{% block title %}Risk Command · Aurora CRM{% endblock %}
+{% block content %}
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h1 class="display-md">De-risk your roadmap</h1>
+      <p class="text-muted">Submit scenario details and optionally connect documents from the catalog. The local model quantifies exposure, recommends mitigations, and surfaces early warnings.</p>
+    </div>
+    <div>
+      <div class="kpi-card">
+        <span class="kpi-label">Local inference</span>
+        <span class="kpi-value">{{ 'Active' if api_ready else 'Offline' }}</span>
+        <p class="text-muted">GPU-accelerated risk synthesis without cloud latency.</p>
+      </div>
+    </div>
+  </div>
+</section>
+
+<section class="panel panel--soft">
+  <div class="hero-grid">
+    <div>
+      <h2 class="display-md">Risk intake</h2>
+      <form id="riskForm" class="form-grid">
+        <label class="form-field">
+          <span class="form-label">Scenario name</span>
+          <input class="form-control" type="text" id="riskScenario" name="scenario" placeholder="Q4 enterprise ramp" required />
+        </label>
+        <label class="form-field">
+          <span class="form-label">Operational background</span>
+          <textarea class="form-control" id="riskBackground" name="background" rows="4" placeholder="Summarize the initiative, dependencies, and signals"></textarea>
+        </label>
+        <label class="form-field">
+          <span class="form-label">Reference documents</span>
+          <select class="form-select" id="riskDocuments" name="riskDocuments" multiple size="6">
+            {% for doc in catalog %}
+              <option value="{{ doc.name }}">{{ doc.name }}</option>
+            {% endfor %}
+          </select>
+        </label>
+        <button class="btn btn--primary btn--full" type="submit">Assess risk</button>
+      </form>
+    </div>
+    <div>
+      <div class="metric-board">
+        <div class="metric-card">
+          <span class="label">Risk score</span>
+          <div class="metric-dial" id="riskDial" data-score="0">
+            <div class="value" id="riskScore">0</div>
+          </div>
+          <p class="text-muted">Composite exposure</p>
+        </div>
+        <div class="metric-card">
+          <span class="label">Priority threat</span>
+          <div class="value" id="riskPriority">—</div>
+          <p class="text-muted">Highest impact risk</p>
+        </div>
+        <div class="metric-card">
+          <span class="label">Early warnings</span>
+          <div class="value" id="warningCount">—</div>
+          <p class="text-muted">Signals requiring monitoring</p>
+        </div>
+      </div>
+      <div class="ai-panel" style="margin-top: 1.5rem;">
+        <p class="text-muted" id="riskSummary">Run an assessment to populate risk narratives.</p>
+        <div class="insight-pillboard" id="riskMatrix"></div>
+        <div class="ai-panel" style="margin-top: 1rem;">
+          <h3 class="eyebrow">Exposure radar</h3>
+          <div id="exposureGrid" class="coverage-bars"></div>
+        </div>
+        <div class="ai-panel" style="margin-top: 1rem;">
+          <h3 class="eyebrow">Early warnings</h3>
+          <ul class="list-stack" id="warningList"></ul>
+        </div>
+      </div>
+    </div>
+  </div>
+</section>
+{% endblock %}
+{% block scripts %}
+  <script src="{{ url_for('static', filename='js/risk.js') }}"></script>
+{% endblock %}
diff --git a/install.bat b/install.bat
new file mode 100644
index 0000000..f866d6a
--- /dev/null
+++ b/install.bat
@@ -0,0 +1,2 @@
+@echo off
+powershell -ExecutionPolicy Bypass -File "%~dp0install.ps1" %*
diff --git a/install.ps1 b/install.ps1
new file mode 100644
index 0000000..71c2b63
--- /dev/null
+++ b/install.ps1
@@ -0,0 +1,4 @@
+param()
+
+$script = Join-Path $PSScriptRoot 'scripts/windows/install.ps1'
+PowerShell -ExecutionPolicy Bypass -File $script @args
diff --git a/requirements.txt b/requirements.txt
new file mode 100644
index 0000000..3f22680
--- /dev/null
+++ b/requirements.txt
@@ -0,0 +1,3 @@
+Flask==3.0.3
+python-dotenv==1.0.1
+requests==2.32.3
diff --git a/run.bat b/run.bat
new file mode 100644
index 0000000..b74eb6f
--- /dev/null
+++ b/run.bat
@@ -0,0 +1,2 @@
+@echo off
+powershell -ExecutionPolicy Bypass -File "%~dp0run.ps1" %*
diff --git a/run.ps1 b/run.ps1
new file mode 100644
index 0000000..23cf335
--- /dev/null
+++ b/run.ps1
@@ -0,0 +1,4 @@
+param()
+
+$script = Join-Path $PSScriptRoot 'scripts/windows/run.ps1'
+PowerShell -ExecutionPolicy Bypass -File $script @args
diff --git a/scripts/windows/common.ps1 b/scripts/windows/common.ps1
new file mode 100644
index 0000000..b25a620
--- /dev/null
+++ b/scripts/windows/common.ps1
@@ -0,0 +1,113 @@
+param()
+
+Set-StrictMode -Version Latest
+$ErrorActionPreference = 'Stop'
+
+function Resolve-ProjectRoot {
+    param()
+
+    $scriptPath = $PSCommandPath
+    if (-not $scriptPath) {
+        return (Get-Location).Path
+    }
+
+    $scriptDirectory = Split-Path -Path $scriptPath -Parent
+
+    $parent = [System.IO.Directory]::GetParent($scriptDirectory)
+    if (-not $parent) {
+        return (Resolve-Path $scriptDirectory).Path
+    }
+
+    $grandParent = [System.IO.Directory]::GetParent($parent.FullName)
+    if (-not $grandParent) {
+        return (Resolve-Path $parent.FullName).Path
+    }
+
+    return (Resolve-Path $grandParent.FullName).Path
+}
+
+function Write-Section {
+    param([string]$Label)
+    Write-Host "`n=== $Label ===" -ForegroundColor Cyan
+}
+
+function Assert-Python {
+    param()
+    $python = Get-Command python3 -ErrorAction SilentlyContinue
+    if (-not $python) {
+        $python = Get-Command python -ErrorAction SilentlyContinue
+    }
+    if (-not $python) {
+        throw "Python 3 was not found. Install it from https://www.python.org/downloads/ and ensure it is on PATH."
+    }
+    return $python.Path
+}
+
+function Ensure-Venv {
+    param(
+        [string]$ProjectRoot,
+        [string]$PythonPath
+    )
+    $venvPath = Join-Path $ProjectRoot '.venv'
+    if (-not (Test-Path $venvPath)) {
+        Write-Host "Creating virtual environment at $venvPath" -ForegroundColor Green
+        & $PythonPath -m venv $venvPath
+    }
+    return $venvPath
+}
+
+function Install-Dependencies {
+    param(
+        [string]$ProjectRoot,
+        [string]$VenvPath
+    )
+
+    $pipPath = Join-Path $VenvPath 'Scripts\pip.exe'
+    if (-not (Test-Path $pipPath)) {
+        $pipPath = Join-Path $VenvPath 'bin/pip'
+    }
+
+    Write-Section 'Installing Python requirements'
+    & $pipPath install --upgrade pip | Out-Host
+    & $pipPath install -r (Join-Path $ProjectRoot 'requirements.txt') | Out-Host
+}
+
+function Load-DotEnv {
+    param(
+        [string]$ProjectRoot
+    )
+    $envFile = Join-Path $ProjectRoot '.env'
+    if (-not (Test-Path $envFile)) {
+        Write-Warning 'No .env file found. Intelligence endpoints will remain offline until OLLAMA_HOST and LLM_MODEL are set.'
+        return
+    }
+    Get-Content $envFile | Where-Object { $_ -match '=' -and $_ -notmatch '^#' } | ForEach-Object {
+        $parts = $_.Split('=', 2)
+        $name = $parts[0].Trim()
+        $value = $parts[1].Trim()
+        if ($name) {
+            Set-Item -Path ("Env:{0}" -f $name) -Value $value
+        }
+    }
+}
+
+function Start-Flask {
+    param(
+        [string]$ProjectRoot,
+        [string]$VenvPath
+    )
+
+    $pythonExe = Join-Path $VenvPath 'Scripts\python.exe'
+    if (-not (Test-Path $pythonExe)) {
+        $pythonExe = Join-Path $VenvPath 'bin/python'
+    }
+
+    Push-Location $ProjectRoot
+    try {
+        Write-Section 'Launching Flask'
+        & $pythonExe -m flask --app app --debug run
+    }
+    finally {
+        Pop-Location
+    }
+}
diff --git a/scripts/windows/install.ps1 b/scripts/windows/install.ps1
new file mode 100644
index 0000000..d45aa4b
--- /dev/null
+++ b/scripts/windows/install.ps1
@@ -0,0 +1,15 @@
+param()
+
+. "$PSScriptRoot/common.ps1"
+
+try {
+    $projectRoot = Resolve-ProjectRoot
+    Write-Section 'Aurora CRM Installer'
+    $pythonPath = Assert-Python
+    $venvPath = Ensure-Venv -ProjectRoot $projectRoot -PythonPath $pythonPath
+    Install-Dependencies -ProjectRoot $projectRoot -VenvPath $venvPath
+    Write-Host "Installation complete." -ForegroundColor Green
+} catch {
+    Write-Error $_
+    exit 1
+}
diff --git a/scripts/windows/run.ps1 b/scripts/windows/run.ps1
new file mode 100644
index 0000000..eaacd7d
--- /dev/null
+++ b/scripts/windows/run.ps1
@@ -0,0 +1,22 @@
+param()
+
+. "$PSScriptRoot/common.ps1"
+
+try {
+    $projectRoot = Resolve-ProjectRoot
+    Write-Section 'Aurora CRM Runner'
+    $pythonPath = Assert-Python
+    $venvPath = Ensure-Venv -ProjectRoot $projectRoot -PythonPath $pythonPath
+    Install-Dependencies -ProjectRoot $projectRoot -VenvPath $venvPath
+    Load-DotEnv -ProjectRoot $projectRoot
+    if (-not $env:LLM_MODEL) {
+        Write-Warning 'LLM_MODEL is missing; set it to an Ollama model name such as llama3.2:latest.'
+    }
+    if (-not $env:OLLAMA_HOST) {
+        Write-Warning 'OLLAMA_HOST is missing; defaulting to http://localhost:11434.'
+    }
+    Start-Flask -ProjectRoot $projectRoot -VenvPath $venvPath
+} catch {
+    Write-Error $_
+    exit 1
+}
diff --git a/uploads/.gitkeep b/uploads/.gitkeep
new file mode 100644
index 0000000..e69de29