AI Trading: Lesson Learned #105: Post-Trade RAG Sync Was Missing

Lesson Learned #105: Post-Trade RAG Sync Was Missing

Date: January 7, 2026
Severity: HIGH
Category: data-integrity, observability, mandate-violation

What Happened

During CEO review on Jan 6, 2026, we discovered that trades were NOT being recorded to Vertex AI RAG or ChromaDB after execution. The pre-session RAG check existed, but NO post-session sync was implemented.

This violated the CLAUDE.md mandate:

"Record every trade and lesson in BOTH ChromaDB AND Vertex AI RAG (MANDATORY)"

Root Cause

• Pre-session RAG check was implemented (reads lessons before trading)
• Post-session RAG sync was NEVER implemented (writes trades after trading)
• JSON backup was working, but not synced to vector databases
• CEO could not query trades via Dialogflow because data wasn't in Vertex AI

Impact

• Trades from Jan 3-6, 2026 were NOT in RAG databases
• Dialogflow queries for trade history returned nothing
• Learning loop was broken (can't learn from trades not recorded)
• Violated CLAUDE.md mandate

Resolution

1. Created scripts/sync_trades_to_rag.py - syncs trades to both Vertex AI and ChromaDB
2. Added "Sync Trades to RAG (Post-Execution)" step to daily-trading.yml
3. Step runs after every trade session, with graceful fallback to JSON

Prevention

• Always verify BOTH pre-flight AND post-flight steps exist for any critical flow
• Test end-to-end data flow: execute trade → record to JSON → sync to RAG → query via Dialogflow
• Add monitoring for RAG corpus size vs JSON trade count

Tags

rag, vertex-ai, chromadb, data-sync, post-execution, mandate-compliance

---

This lesson was auto-published from our AI Trading repository.

More lessons: rag_knowledge/lessons_learned