Title: Chatbot Platform for BL and Others | Brokk & Sindre Description: BL needed a scalable chatbot platform capable of serving multiple organizations safely, each with different knowledge bases, boundaries and requirements. Canonical: https://brokk-sindre.dk/en/cases/bl-chatbot - [Home](/en) - [Cases](/en/cases) - Bl chatbot # Chatbot Platform for BL and Others BL needed a scalable chatbot platform capable of serving multiple organizations safely, each with different knowledge bases, boundaries and requirements. By Mikkel Krogsholm · March 15, 2024 An AI service becomes useful when its sources, boundaries, responsibilities and ongoing maintenance are clear. The model is only one part of the decision. - Platform serving BL and Esmark and previously Junkbusters - Full RAG pipeline with organization-specific knowledge bases ## Background BL is the umbrella organization for the Danish social housing sector. It needed an AI chatbot that could answer questions about legislation, maintenance and tenant relations and also serve other organizations with their own knowledge. ## My approach I designed and built Huginn & Muninn, named after Odin’s two ravens. Each organization gets its own knowledge base and configuration while sharing the underlying infrastructure. ### Tech stack - **Flowise** for conversational AI flows - **n8n** for workflow automation and integration - **[Qdrant](https://qdrant.tech)** vector database for semantic search - **PostgreSQL** for metadata and configuration ## Results The platform has run across multiple organizations, handling inquiries on everything from legislation to practical maintenance. Junkbusters is no longer a client. That operating history made source control, isolation and ongoing ownership concrete design requirements rather than theoretical concerns. ## Do you face a consequential AI decision? Write about the decision, or ask about a leadership briefing or talk.