Optiak, a startup building what it calls a modular operating system for enterprise AI, has come out of stealth with a €4m pre-seed round, about $4.7m, led by Market One Capital, Next Tier Ventures, EA Ventures Plug and Play EMEA Fund and Mission. The company, the founders say, is trying to solve a problem most large organisations have walked into rather than chosen: a sprawl of AI tools that nobody is governing as a single system.
The pitch is an intermediary layer. Optiak sits between any AI application a company runs, a chatbot, an agent, a workflow, and the models underneath, giving a single point where security, governance, memory, observability, and cost optimisation are configured once rather than rebuilt app by app.
In practice, the company says, that means corporate policies can block an IP or security risk before an employee sends anything to a third-party model, and the platform can then route the request to whichever permitted model answers best for the lowest cost.
The problem it is aimed at has a name and some numbers behind it. “Shadow AI,” employees reaching for tools without IT oversight, has produced internal data leaks at 46% of organisations, according to Cisco’s 2025 research, and unsanctioned AI now accounts for roughly one in five breaches in IBM’s 2025 breach report.
