Kaizen

Kaizen use AI to analyse developer impact and offer an actionable theory of improvement. Demo video.

What is Kaizen?

Kaizen helps engineering leaders at small to medium-sized companies understand team productivity, evaluate AI adoption impact, and get specific recommendations for improvement. Unlike traditional dashboards that show metrics without context, Kaizen uses AI to provide a "theory of improvement"—telling you not just what's happening, but what to do about it.

Who is this for?

Engineering leaders at companies with fewer than 500 engineers who want to:

• Identify high-impact contributors and emerging talent
• Understand where their team stands relative to industry benchmarks
• Measure the productivity impact of AI tools (similar to METR's methodology)
• Get actionable recommendations instead of just dashboards

Core Features

Individual Developer Profiles

Each contributor gets a radar chart across 9 dimensions:

• Code Velocity - Commit frequency and volume
• Review Quality - PR review depth and engagement
• Code Stability - Bug rate in contributed code
• Collaboration - Cross-team work patterns
• Incident Response - Production issue resolution speed
• Feature Delivery - Completed tickets and shipped work
• Documentation - Knowledge sharing contributions
• Testing Rigor - Test coverage improvements
• Availability - Consistent contribution patterns

AI-Powered Insights

Powered by NVIDIA NIM (NVIDIA Inference Microservices):

• Chat interface: openai/gpt-oss-20b - Smaller, faster model for interactive queries
• Executive reports: openai/gpt-oss-120b - Larger model for comprehensive analysis

The system analyzes your metrics and generates:

• Diagnosis - Where your team underperforms relative to benchmarks
• Theory of Improvement - Research-backed practices that correlate with better outcomes
• Recommended Projects - Specific initiatives ranked by expected impact

Example: "Your deployment frequency is 2x/week vs. industry median of 8x/week. Teams that improved this metric typically adopted trunk-based development and automated testing. Priority: Implement CI/CD for your 3 highest-traffic services."

AI Impact Measurement

Track productivity changes after AI tool adoption:

• Before/after metric comparison
• Individual variation in AI effectiveness
• ROI analysis on AI tooling investments

Data Sources

Kaizen integrates with:

• GitHub - Code commits, PRs, reviews, file changes
• Linear - Ticket completion and feature delivery
• Datadog - Incident response, deployment frequency, error rates

How It Works

1. Connect - Authenticate with GitHub, Linear, and Datadog
2. Sync - Pull 90 days of historical data
3. Analyze - AI calculates dimension scores and identifies patterns
4. Review - View team dashboard and individual radar charts
5. Act - Receive weekly recommendations via email

Key Differentiators

Actionable vs. Observational - Most tools show you lines on a graph. Kaizen tells you which projects to prioritize.

Individual + Team - Understand both aggregate team health and individual contribution patterns.

Research-Grounded - Synthesizes DORA, SPACE, and Accelerate research into specific recommendations for your context.

AI-Native - Built for the era where AI tools are part of every developer's workflow.

Quick Start

npm install
cp .env.example .env.local
# Add API keys for GitHub, Linear, Datadog, Resend, Supabase, NVIDIA
# Optional: set AGENT_MODEL_NAME to override the default chat model (openai/gpt-oss-20b)
npm run dev

Get your free NVIDIA API key at https://build.nvidia.com/

Visit http://localhost:3000 and trigger your first sync.

Adding additional repositories

• Click Add Repo in the dashboard header (or use the quick guidance card) and either pick from your GitHub repositories list or enter owner/repo manually
• Authorize GitHub when prompted. We request repo (read) + read:user scopes and never store the access token server-side.
• Once connected, pick any repository you have read access to. The new repository appears in the selector after the initial sync (usually under a minute).

---

Note: This tool is designed for team improvement, not individual performance evaluation. Use responsibly.

Supabase Auth & GitHub OAuth

1. Add to .env.local:

NEXT_PUBLIC_SUPABASE_URL=your_project_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_anon_key

2. In Supabase Dashboard → Authentication → Providers:

   • Enable Google (optional) and paste your OAuth Client ID/Secret if you want Google sign-in.
   • Enable GitHub. Create a GitHub OAuth app (Developer settings → OAuth Apps) with redirect URI https://<your-project-ref>.supabase.co/auth/v1/callback. Paste the Client ID/Secret back into Supabase.
   • Add additional scopes repo read:user so we can list private repositories.
   • Toggle “Save provider refresh tokens” so Supabase exposes the GitHub access token in the session.

3. In the app, sign in with GitHub from the user menu when prompted. The OAuth callback is handled at /auth/callback.

4. A public.users row is created automatically for each authenticated user.