Pre-submission Checklist
Area
Search / filters
Problem or Use Case
Memos currently only supports strict keyword exact-match search. Users can only find memos by typing the exact words originally recorded. This creates a poor user experience for memory retrieval scenarios. In most cases, users only remember the general meaning and context of their past notes, but not the precise wording. For example, a memo about “drinking warm water in the morning” cannot be retrieved by searching “morning hydration habit”; a memo about “weekly Sunday work summary” cannot be matched by searching “weekly work review”. Existing keyword search fails to retrieve valid content, resulting in missing records and low retrieval efficiency.
Who Would Use This?
All Memos users, especially long-term users who have accumulated a large number of memos. It is highly suitable for personal knowledge management practitioners, daily note-takers, self-hosted users, and users who frequently retrieve historical records. Semantic search perfectly fits human memory habits: remembering the idea rather than the exact words, greatly improving the usability of Memos for long-term memory and knowledge accumulation.
Proposed Solution
Add native AI semantic search (meaning-based search) capability to Memos, replacing single keyword matching with semantic similarity matching.
Core features:
- Support natural language fuzzy search: match memos with similar semantics even without overlapping keywords.
- Compatible with the original keyword search: intelligently adapt or manually switch between exact keyword search and semantic search.
- Support local lightweight embedding models and third-party AI interfaces such as Ollama/OpenAI, adapting to self-hosted and privacy-first deployment scenarios.
- Sort results by semantic similarity to display the most relevant memos first.
- Multi-language Model Compatibility: Support mainstream universal multi-language embedding models to serve global users with diverse writing habits. Most default embedding models are heavily English-optimized, which results in inaccurate semantic matching for non-English daily conversational notes and short sentences. Better multi-language support will greatly improve search accuracy for non-Latin script user scenarios, including everyday informal notes written in Chinese and other common languages.
Alternatives Considered
No response
Compatibility and Migration
No response
Examples
No response
Additional Context
No response
Pre-submission Checklist
Area
Search / filters
Problem or Use Case
Memos currently only supports strict keyword exact-match search. Users can only find memos by typing the exact words originally recorded. This creates a poor user experience for memory retrieval scenarios. In most cases, users only remember the general meaning and context of their past notes, but not the precise wording. For example, a memo about “drinking warm water in the morning” cannot be retrieved by searching “morning hydration habit”; a memo about “weekly Sunday work summary” cannot be matched by searching “weekly work review”. Existing keyword search fails to retrieve valid content, resulting in missing records and low retrieval efficiency.
Who Would Use This?
All Memos users, especially long-term users who have accumulated a large number of memos. It is highly suitable for personal knowledge management practitioners, daily note-takers, self-hosted users, and users who frequently retrieve historical records. Semantic search perfectly fits human memory habits: remembering the idea rather than the exact words, greatly improving the usability of Memos for long-term memory and knowledge accumulation.
Proposed Solution
Add native AI semantic search (meaning-based search) capability to Memos, replacing single keyword matching with semantic similarity matching.
Core features:
Alternatives Considered
No response
Compatibility and Migration
No response
Examples
No response
Additional Context
No response