Captain

Knowledge retrieval and enterprise search infrastructure is a massive and fast-growing B2B market as every enterprise scrambles to make internal data AI-queryable. ICP is reasonably clear (companies with large unstructured file stores needing accurate retrieval). API-first infrastructure positioning enables broad monetization via usage-based pricing.

Edgar Babajanyan (CTO): Purdue Network Engineering (Dean's List), Reality Interactive AI Software Engineer for 2.5 years — built production RAG for Confluence documentation, AI agents for customer service, and OCR models (directly what Captain does), Purdue NLP researcher (published paper on LLM annotation tools), Weaviate vector DB certified, Autonomous Robotics Club Head of SW Engineering (software used by Google/Amazon/Unitree). Direct production RAG engineering experience before founding. Lewis Polansky (CEO): Purdue Finance (Honors), self-taught full-stack engineer, founded CyberSpace cybersecurity club (4 years, grew to 3 locations, trained national CyberPatriot semifinalists), Congressional App Challenge winner recognized by House of Representatives, Eagle Scout. Applied to YC 3 times with 4 pivots — real grit and iteration. Not elite schools and no prior exits, but Edgar's production RAG background is directly domain-relevant.

This space is extremely crowded: LlamaIndex, LangChain, Cohere Embed, Pinecone, Weaviate, and dozens of RAG-focused startups all compete here. OpenAI and Anthropic are building retrieval natively into their products. The differentiation claim (higher accuracy) is meaningful if proven but not yet a durable moat — any well-resourced competitor can claim similar improvements.