Insight for: v3: semantic query with embeddings

Graphify's query mechanism, evolving from keyword-based BFS to embedding-based semantic search.

Analyzed: Apr 8, 2026

This issue outlines a critical upgrade for Graphify's query functionality, transitioning from keyword-based BFS to embedding-based semantic search. The current limitation, requiring exact string matches, severely restricts its utility as an 'understanding tool.' The proposed shift to semantic search, leveraging local embedding backends like `sentence-transformers` or optional API-based solutions, directly addresses this pain point. By enabling queries based on conceptual meaning rather than literal keywords, Graphify significantly enhances its value proposition for developers navigating complex codebases. This evolution positions Graphify as a more intelligent, intuitive tool for code comprehension and analysis. For B2B SaaS, this feature is a strong differentiator, moving the product beyond basic search to advanced semantic understanding, which is crucial for enterprise-level code intelligence and modernization efforts.

BFS keyword matching grep with graph traversal embedding-based semantic search find concepts by meaning `sentence-transformers` `all-MiniLM-L6-v2` OpenAI embeddings API nomic-embed via ollama graph build embed every node label + source context store vectors in `graph.json` query embedding cosine similarity top-k hits `semantically_similar_to` edge detection LLM Node similarity graph visualization

GitHub Issue

Parent Entity

v3: semantic query with embeddings

State: Open