Skip to content
Specor

Semantic Search

As your product model grows, search is how you find your way around it. Specor searches across every node in your graph — Features, Capabilities, Rules, Flows, Personas, Metrics, and more — and combines two kinds of matching so you find things whether or not you remember the exact wording.

Keyword and semantic matching

Specor uses hybrid search:

  • Keyword matching finds nodes whose titles or descriptions literally contain your terms
  • Semantic matching finds nodes that are about what you searched for, even when the words differ — so a search for “abandoned cart” can surface a Feature titled “Checkout drop-off recovery”

Results are ranked by relevance across both signals, so the most useful node tends to come first regardless of exact phrasing.

Search spans the whole product graph on the active branch:

  • Nodes by title and description, filtered by type (e.g. only Features, only Rules)
  • The relationships around a match, so you can jump from a result into its neighbourhood in the graph

Open search from the app and type your query. Pick a result to jump straight to that node’s detail view, where you can see its rules, acceptance criteria, flows, and connected nodes.

Semantic search is included on the Team plan and above.

Search and the graph

Search complements the graph view rather than replacing it. Use search when you know what you’re looking for and want to jump there directly; use the graph when you want to explore how things connect. Both read the same live product model.

Tips

  • Search by intent, not just labels. Because matching is semantic, describing the behaviour you’re after often works even if you don’t know the node’s exact name.
  • Filter by type to cut noise when you know you’re after a Rule or a Metric specifically.
  • Your AI tools search too. Over the MCP server, assistants like Cursor and Claude can search your product graph the same way — see search_features and related tools.