Key Capabilities
- Search across all indexed meetings from one workspace
- Hybrid retrieval combines SQLite FTS5 lexical search with local embeddings
- Ask questions across the whole meeting library with source citations
- Filter by meeting, date range, and evidence type
- Index transcript passages, summaries, screenshot analysis, and meeting metadata locally
Your meeting history becomes queryable
Instead of opening meetings one by one, search the whole library for a customer name, technical decision, deadline, risk, or recurring topic. Results preserve the source meeting and evidence location so you can open the original context.
Local hybrid RAG
RecapFox combines lexical retrieval with local semantic embeddings, then resolves the strongest current meeting evidence before generating an answer. If semantic embeddings are unavailable, full-text retrieval can continue independently.
Learn the workflow
- How to Take Notes During an Architecture Review — Capture architecture decisions, trade-offs, constraints, rejected alternatives, benchmark evidence, and the diagrams that make the system understandable later.
- How to Take Notes During Client Discovery Calls — Capture problems, current workflow, constraints, decision criteria, objections, commitments, and exact customer language without losing the source context.
- A 10-Minute Workflow for Reviewing a Meeting After It Ends — Verify the summary, check decisions and action items, attach visual evidence, resolve ambiguity, and make the meeting searchable before context disappears.
- How to Build a Meeting Knowledge Base You Can Actually Search — Turn a growing archive of transcripts, summaries, and screenshots into a useful retrieval system with clear evidence, good meeting titles, and hybrid search.
Local-first architecture: Meeting-content transcription, language-model reasoning, embeddings, and retrieval are designed to run locally rather than requiring a hosted inference API.