Monster AI: Back Market's Odyssey to a Custom Generative AI Platform
summary
Back Market describes building Monster, an in‑house generative AI platform on GCP using Chainlit/LangChain, FastAPI, Cloud Run and vector stores, with optional connectors to Vertex AI or OpenAI. It integrates Okta authentication, query/feedback logging, input PII handling (anonymize/edit/continue), LLM‑based CI tests, fine‑tuning from user feedback, and RAG over Confluence and internal docs; deployment serves ~200 daily users across 700 employees and supports ~150 support agents with human‑in‑the‑loop review.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 10:45Security analysis and risks
- 17:26Expansion and use cases
- 22:59Key learnings
- 25:25Monster AI demo
- 29:58Q&A
related talks
22mLearning to Apply Generative AI at Enterprise Level
28mUnlocking the power of GenAI
16mTransform customer experience with an array of AI agents
24mHow to create and scale AI Products, MLOPs and Generative AI
07mBuild Generative AI Apps for Enterprises with Cloud
34mUnlock the full potential of Generative AI with Agent Platforms
29mCustomizing genAI applications for your business
09mGen AI for regulated industries