Back to Work
Adopted by a 50-person product org
PlanningSync
Built and deployed an AI-native Shape Up planning platform used across a 20-engineer, 30-person product org. Features LLM scope-coverage analysis, hill charts, an MCP server, two-way Jira/GitHub sync, and automated Slack digests.
TypeScript · React · MCP Server · LLM Integration
The Problem
At SalesAI, product delivery visibility depended on manual status updates. Leadership couldn't see where each Shape Up bet actually stood, or where scope was quietly drifting from the pitch.
Architecture & Approach
- Built hill charts that update automatically from GitHub PRs and commits plus two-way Jira sync, so engineers don't maintain status by hand.
- Engineered a structured-output LLM analyzer that audits planned work against Confluence pitches, flagging uncovered concepts and rabbit-hole risks by meaning rather than keywords.
- Added confidence-scored estimation suggestions based on historical work.
- Exposed planning data to AI assistants through an MCP server.
- Made opinionated product calls: explainable tag suggestions, auto-fading stale bugs, noise-free stakeholder views, and automated Slack digests.
Measurable Impact
- Adopted daily across 20 engineers and a 30-person product organization.
- Became the single source of truth in weekly all-hands reviews, presenting live hill charts of all open Shape Up bets.