How an RTE and a PM co-built an AI-native practice in real time
summary
Pluxee teams race to be PI-ready in under three weeks and build two independent agents (Rovo linked to Jira/Confluence and a Copilot‑refined prompt flow) to prepare a PRD and discovery artifacts; BMAD proves too slow and noisy so Rovo becomes primary, producing an MVP chat assistant inside the EVA app that works in tests, reduces FAQ bounce and aims for 15% contact reduction and 70% FAQ accuracy, validated by dev/security sign‑offs and close beta users via two agents in conversation.
AI-written summary · about
