AI that does the work.

Chat assistants answer; Onky acts. It executes across the systems you run: updates the CRM, chases the invoice, posts the summary to Teams, and builds the tools it is missing. Every action passes through Onky's policy engine, the Kernel, with a full audit trail, and your data never reaches a third-party AI lab.

It builds it. The Kernel, our policy engine, saves the task with its context, builds the missing tool, and runs it through policy checks. The original task then resumes and completes, and the new tool becomes a governed capability your whole organisation can reuse.

The company brain is Onky's governed picture of how your organisation works: processes, owners, handoffs, and tools, assembled from the systems you connect and kept current by the work Onky executes. Access is role-based, audited, and under your retention control. It is what lets agents act with context instead of guesses.

Yes, exclusively. Onky runs open-weight models inside EU infrastructure it controls, every call logged and policy-checked by the Kernel, our policy engine. Self-hosting the models is what makes the guarantees possible: your data stays confidential, your IP is protected, and costs stay controllable with no per-token API bill. It is also why Onky can run on-premise and air-gapped, where proprietary-API models can't follow.

On Onky's EU cloud infrastructure (europe-west1), kept in region. The models are open-source and run inside that same infrastructure: your data never reaches a third-party AI lab and is never used to train models. Access is role-based and logged, deletion is available on request, and Onky signs a DPA as your data processor at pilot signing. If you need full isolation, on-premise and air-gapped deployments are available.

Today Onky runs on its managed EU cloud (europe-west1). Private VPC is next on the roadmap, and on-premise and air-gapped deployments, with Onky's policy engine, the Kernel, fully inside your own perimeter, are available on request. Book a demo and we will scope it with you.

Onky is built around what the Act asks operators to demonstrate: a complete audit trail, transparency over automated decisions, and human oversight through your policy gates. Every build, deployment, and model call runs through Onky's policy engine, the Kernel, and is logged end to end. Onky does not certify your compliance; it gives you the governed architecture and the evidence.

Shadow AI is what happens when employees use AI tools outside governed channels. Onky closes that path by being the governed channel: every model call, build, and deployment runs through the Kernel, our policy engine, and every connector is governed by default via MCP. Nothing operates outside your policies or your audit trail.

Access is invitation-only today: we are running private pilots. It starts with a 30-minute call to understand your needs. Then, during onboarding, Onky discovers which tools your teams are missing and which agents and automations they need. From there you just start using it: whenever something new pops up, Onky builds it, so your team stops losing time to repetitive tasks.