Blitzy Inc. secured $200 million in funding in May, reaching a valuation of $1.4 billion. The platform enables the parallel operation of thousands of autonomous coding agents.
The company utilizes Neo4j graph databases to provide agents with codebase context. By reverse-engineering environments, the platform maps code modules, files, and functions as interconnected entities.
Traditional search methods limit agents to approximately 200,000 to 300,000 tokens, or 20,000 to 30,000 lines of code. Graph structures allow agents to navigate 100-million-line codebases without losing information.
The platform incorporates human oversight, requiring approval of an Agent Action Plan before execution. In June, the system achieved an 84.95% score on the SWE-Bench Pro benchmark.
Agents use the Cypher query language to interact with the database. Because the language requires precise queries, it prevents agents from operating on incorrect or hallucinated data.
