PitWall AI

What Inspired Us to Build PitWall AI : Modern software teams move fast, but coordination often breaks down before code review even starts. Pull requests get reviewed in isolation, Jira tickets go stale, and important context lives across tools and people’s heads.

We were inspired by Formula 1 race engineering, where drivers focus purely on driving while the Pit Wall continuously monitors telemetry to anticipate risks and intervene early. That model maps surprisingly well to software teams: developers code, while tech leads and reviewers need early warning signals about risk, scope, and coordination gaps.

Existing tools focus heavily on code quality and style, but very few address process risk and change impact before review begins. PitWall AI was built to fill that gap — acting as a race engineer for software teams, not a code critic.

What We Learned While Building This Project : Building PitWall AI taught us several important lessons:

AI is most effective as an advisor, not a decider Using Atlassian Rovo to generate “Pre-Race Checks” instead of approvals helped keep humans in control while still benefiting from AI reasoning.

UI language shapes user trust Framing feedback as “signals”, “flags”, and “warnings” (instead of errors or judgments) dramatically improved how the tool felt to use.

Scope control is a technical skill Intentionally limiting the MVP to a Jira Issue Panel helped us ship something cohesive rather than an unfinished multi-product integration.

Challenges We Faced During Development : Understanding Forge’s execution model Forge does not behave like a traditional backend or frontend environment. Learning how resolvers, UI modules, tunneling, and deployment interact took time and experimentation.

Balancing AI and deterministic logic Early versions leaned too heavily on AI. We had to redesign the system so AI explains and augments risk rather than defining it outright.

Context extraction from Jira issues Jira issues vary wildly in quality and structure. Designing logic that works across sparse and detailed tickets was a real challenge.

Communicating intent clearly We had to be explicit that PitWall AI is not a code reviewer, but a coordination and risk-awareness tool — both in the UI and in documentation.

Built With

• forge
• javascript
• jira
• react
• rovo
• tailwind