How Tiger Data Built a Production AI Slack Bot with Pydantic AI and Logfire

Tiger Data, a fully remote company, has developed Tiger Agent, an AI Slack bot designed to manage extensive conversations and provide relevant context efficiently. This initiative arose from the challenge of maintaining coherent communication within a growing company using Slack as its primary platform. To build the Tiger Agent, the company selected Pydantic AI for its LLM orchestration and Logfire for observability, valuing their ability to provide abstraction without vendor lock-in and comprehensive observability. The integration of multiple MCP servers allows the agent to access various data sources, enhancing its capability to address questions about Tiger Data's products and customer issues. The use of Pydantic AI facilitates seamless switching between LLM providers and automates retry logic, while Logfire's structured data approach and SQL-based querying improve traceability and debugging. This combination enables Tiger Data to maintain high engineering quality and reliability in their AI systems without being locked into a specific framework, ultimately allowing for enhanced visibility and reduced development complexity.