Semantic search uses AI embeddings to find jobs and contacts by meaning, not just keywords.
How it works:
- Go to any search field in Orbit
- Type a natural description like "roles focused on distributed systems" or "contacts who work in fintech"
- Orbit uses AI to compare your query against all your jobs and contacts
- Results are ranked by semantic similarity, not keyword matching
What gets searched:
- Job titles, companies, descriptions, and notes
- Contact names, titles, companies, and notes
- Cross-matching: find contacts relevant to a specific job, or jobs matching your resume profile
Requirements:
- Ultra plan
- OpenAI API key configured in Settings (embeddings use text-embedding-3-small)
How embeddings work:
When you save or update a job or contact, Orbit generates an embedding (a numerical representation of the text) and stores it in the database. Searches compare your query embedding against all stored embeddings using cosine similarity.
Backfill existing data:
If you already have jobs and contacts, go to Settings to trigger a backfill that generates embeddings for all existing records.
Tip: Semantic search is most powerful when your job descriptions and notes are detailed. The more context you add, the better the matching.