source: techcrunch ai: zerodrift raises $10m to protect ai models from themselves
level: business
zerodrift raised $10 million in a seed round led by a16z speedrun, reign ventures, pitchdrive ventures, and u&i ventures. the company builds a compliance layer that sits between ai models and end users. it flags messages that might break rules like soc 2 or gdpr, then uses a large language model to rewrite them safely. the ceo, kumesh aroomoogan, says the system first uses deterministic programs to spot violations, then calls on llms only for the rewrite. this approach aims to be faster and more reliable than relying on a single ai model to police itself.
the service targets ai chatbots that face consumers, where wrong answers can cause real harm. but aroomoogan sees a bigger market in automated messages that humans never see. as ai spreads, the need for such guardrails will grow. the fundraising was quick, closing in three weeks and oversubscribed by three times. aroomoogan credited a16z for helping structure the round. the company argues its architecture gives it an edge over built-in safety features from labs like openai and anthropic, which are often already part of the underlying system.
zerodrift's method separates detection from correction. conventional programs catch problems deterministically, so the llm only handles rewriting. this split can lower latency and improve reliability compared to end-to-end ai moderation. the company enters a small but expanding market for ai governance tools. enterprises are increasingly worried about compliance as they deploy ai, and dual-model setups are becoming common. zerodrift focuses entirely on the second, corrective model, offering a specialized service for a growing pain point.
why it matters: it shows a practical way to make ai outputs safer without slowing down systems, which matters for any business putting ai in front of customers.
source: techcrunch ai: zerodrift raises $10m to protect ai models from themselves