Inside our effort to improve the Mintlify assistant

Efforts to improve the Mintlify assistant focused on rebuilding the feedback pipeline, transitioning conversation data to ClickHouse, and categorizing negative interactions. The analysis revealed that search quality was the assistant's main weakness, although most responses were strong. To address these issues, Mintlify updated its server to map feedback events to original conversation threads, enabling effective querying of negative feedback. The team categorized feedback into eight types using a large language model (LLM) and highlighted search as a key area for improvement. Despite a model upgrade, feedback consistency remained unchanged, leading to the expansion of the insights tab for better understanding customer concerns. UI improvements included better navigation through conversation threads and more stable response loading. Users are encouraged to provide feedback or join the Mintlify team to further enhance the assistant.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.