An engineer posted to Hacker News this week alleging that a startup founder is publicly claiming credit for a RAG (Retrieval-Augmented Generation) architecture the engineer spent two years building, presenting it as the company's flagship AI work. The post, filed under "Ask HN," names neither the company nor either party. The specifics of the employment or contractor arrangement were not disclosed.
The dispute centers on work that carries real commercial weight. A two-year production RAG build — the kind companies use to ground LLM outputs in proprietary knowledge bases, cutting hallucinations and enabling document-level retrieval — can anchor a sales pitch or a <a href="/news/2026-03-14-captain-yc-w26-launches-automated-rag-platform-for-enterprise-ai-agents">Series A narrative</a>. Founders who can credibly claim early, hands-on RAG expertise have used that story to close enterprise deals. That's what makes credit over this kind of system contentious: it's a business asset, not an academic footnote.
The thread's top responses pointed the engineer toward their offer letter or contractor agreement, specifically the IP assignment clauses. The legal outcome turns on jurisdiction and employment status. Work-for-hire doctrine treats employee-created IP differently from contractor work, and some states require an explicit written assignment rather than assuming ownership transfers automatically. Without the contract terms on the table, the thread could only take the analysis so far.
The bluntest advice in the replies: get IP terms in writing before the work starts. Startups move fast and formalize later — and that's how engineers end up posting to HN two years in.