Koralis AI builds Kovac, software that checks requests to your data against your policy, for organisations putting AI to work on data they must protect. Kovac is in beta.
What it does in our demos today: an AI step is refused before it reads anything your policy doesn't allow, and the refusal says why in plain words. People and AI tools are held to the same rules. Columns you mark as sensitive are tokenised before any AI or webhook call. Refusals and AI calls go into a signed, chained record that can be exported and checked offline. Its checks are deterministic: Kovac is not an AI model and does not supply one; it governs the AI others provide.
In a BridgeAI project we would be the governance partner. The deployer decides what the AI developer's system may see, sensitive fields stay out of the model, and the compliance team has a record it can check. That fits the Assurance and Compliance strand best, and professional and financial services firms in particular.
We can't lead, as we supply no AI or ML solution of our own. We are looking for an AI SME and a deployer building a bid, to join as their third partner.
Founded by Dylan Banks (board-level AI strategy) and Marko Krznaric (former Palantir Forward Deployed Engineer). Koralis AI Limited, company number 16558134.
Social media
Areas of activity
Industrial Sector
Digital and TechnologiesFinancial ServicesProfessional and Business Services
We'll join an AI SME and a deployer as third partner. Kovac checks the AI's requests to the deployer's data against its policy and records refusals. In beta.