VizuaraAI/ramco-journey-model

Ramco PO Chatbot · v9

Interactive test environment for the Ramco Purchase Order chatbot. Bot v9 · 77.6% eval pass (45/58) · journey-driven, deterministic decisions, LLM only for NL understanding.

v9 closes a structural gap from v1–v8: when the user walks a sub-screen (Terms, TCD, PR-Cov, SO-Cov, Quality, Notes, Schedule), the sub-screen's own TRANS task now fires alongside the main commit — writing the orphan child tables (po_paytm_doclevel_detail, po_potcd_doclevel_detail, po_poqly_quality_detail, …) that were previously never touched. All 8 new sub-screen-commit eval cases pass; zero regressions vs v8.

Set the GEMINI_API_KEY secret in this Space for the bot to function.

Source: https://github.com/VizuaraAI/ramco-journey-model