AI Pit Strategist

AI-powered Rovo Agent bringing Formula 1 precision to Agile teams—analyze sprint health, auto-triage issues, and generate retrospectives with racing-inspired insights.

Comment

💡 Inspiration

Software development is a high-speed race. Engineering managers and scrum masters are under immense pressure to deliver, but they are often stuck in the pits, buried under a mountain of disjointed tools. They constantly switch contexts between Jira for tasks, Slack for standups, GitHub for code reviews, and meetings.

This fragmentation forces them to "drive blind"—guessing their velocity, ignoring burnout warnings, and missing the signals of upcoming blockers.

We were inspired by the F1 Pit Wall: the nerve center where millions of data points are analyzed in real-time to make split-second strategic decisions. We asked: What if we could unify all these signals and give every Agile Team their own AI Race Strategist?

🏁 What it does

AI Pit Strategist is a Unified Mission Control for high-performance software teams. It aggregates telemetry from Jira, GitHub, and Slack into a single, F1-inspired dashboard.

  • 🏎️ Race Telemetry (Sprint Health):
    • Live Sprint Probability: Forget static burn-down charts. We calculate the live probability of finishing on time (displayed as "Fuel Levels" and "Tire Health").
    • Code Health Radar: A visual radar chart mapping Code Quality, Test Coverage, and PR Velocity directly from your GitHub repo.
  • 📻 Pit Radio (AI Agent):
    • A context-aware Rovo Agent that lives on your dashboard.
    • "Auto-Response" Actions: One-click generation of replies to stale PRs, Jira issues, and comments using the agent's context.
    • Natural Language Issue Creation: "Crew, we need to fix the login bug on staging." -> Automatically creates a detailed Jira ticket with labels and priority.
  • ⚡ Playbooks & Automation:
    • Rovo Playbooks: Configurable rules (e.g., "If PR is stale for 3 days -> Slack the team") that run automatically to keep the race track clear.
    • Meeting Action Extractor: Parses meeting notes to automatically create follow-up tasks.

⚙️ How we built it

We engineered a high-performance Hybrid Architecture to bring the F1 experience to life:

  1. Frontend (The Cockpit):

    • Built with React, Vite, and TailwindCSS.
    • Custom Design System: We created a "Mission Control" aesthetic with dark-mode coding, glassmorphism panels, and neon status indicators (Green/Yellow/Red flags) defined in our CSS variables.
    • Visualization: Heavy use of visual widgets (SprintHealthIndicator, LapProgressBar, CodeHealthRadar) instead of text lists.

  2. Backend (The Engine):

    • Hosted on Vercel to handle computationally expensive AI tasks and 3rd-party webhooks.
    • Supabase (PostgreSQL): Stores team "telemetry" data and user preferences.
    • Drizzle ORM: Ensures type-safe database interactions across our TypeScript codebase.

  3. Integrations (The Sensors):

    • Jira Software: Deep two-way sync for issues and sprints.
    • GitHub: Webhooks for PR events and code analysis.
    • Slack: Real-time notifications and "Pit Radio" broadcasts.

🔧 Challenges we ran into

  • Unifying Auth: Handling both Jira (Atlassian Connect/Forge) and GitHub OAuth flows in a single session was complex. We built a robust provider-switching mechanism to allow valid tokens for both platforms simultaneously.
  • Visualizing "Health": Translating raw data (ticket count) into a meaningful "Sprint Health" score required iterating on our probability algorithms to avoid false alarms.
  • Real-time Context: Feeding the Rovo Agent the right context (latest PR, specific Jira comment) without exceeding token limits required careful prompt engineering and context pruning.

🏆 Accomplishments that we're proud of

  • The "Pit Radio" Experience: It genuinely feels like talking to a Race Engineer. The ability to click "Auto PR Response" and watch the AI draft a code-aware reply is magical.
  • Design Polish: We moved away from the standard "corporate SaaS" look to something that feels exciting and premium, like a true dedicated dashboard.
  • Hybrid Power: Successfully marrying the secure context of Atlassian Forge with the flexibility of a Vercel backend.

🚀 What's next for AI Pit Strategist

  • Biometric Integration: (Experimental) Using wearable data to measure developer "driver fatigue" and suggest breaks.
  • Multi-Team Races: Gamification where squads "race" against each other on velocity and quality metrics.

Built With

Share this project:

Updates