Blitzy bets on knowledge graphs to power autonomous coding agents

Blitzy uses a Neo4j graph database to give coding agents the context needed to safely modify large enterprise codebases. The company raised $200 million at a $1.4 billion valuation and reported an 84.95% score on SWE-Bench Pro.

• Blitzy raised $200M at a $1.4B valuation in May
• Agents query the code graph via Neo4j and Cypher
• Effective agent context is 200,000–300,000 tokens
• SWE-Bench Pro score reached 84.95% in June