Notes from Odokai

The ICO has opened a call for evidence on agentic AI, closing 20 November. What it covers, six checks before an agent touches personal data, and what does not change yet.

Five stages, three approaches, one honest warning about the traps. How SMEs cut invoice handling time by 70 to 90 percent, and what the first working version looks like.

Both are good. The right answer depends on where your work happens, who is using it, and what the data is. Three questions settle it.

The honest answer: free ChatGPT, no. Enterprise, mostly. And for regulated work, "mostly" is not good enough. What actually happens to your data, and the safer routes.

The four categories of AI tool UK small businesses actually choose between, the data checks to run before signing up, and a test that settles what is worth the money.

Per-seat assistants quote a number you can see. Agent platforms quote a number you have to work for. Both, with real figures, and the total-cost maths to run first.

Sovereignty, provenance, and proof in one platform. Runs in your cloud account or ours, on any model, with a provenance trail on every output. Already in use for regulated work.

The AI platform for SME productivity. Every person gets a team of shared agents, inside your own environment. Two videos show what that looks like.

UK SMEs spend about £23K a year on AI tools. The true cost is closer to £57K–£71K once overlap, waste, and compliance gaps are counted.

Your team spends hours on work that should take minutes. We find that work, build the tooling, and put it live on a platform your team already knows how to use.

Coding agents invert the assumption CI was built on. ACIG is our answer: a tiered, parallel critic fleet that runs cheap, structured checks before involving a human or frontier model.

When the platform owns security, identity, and interoperability, teams can create and discard capability safely. Introducing ephemeral software and why the future is not more apps.

Base44 proved AI app building can be simple. Odokai delivers similar speed for teams that need private infrastructure, auditability, and permanent ownership.

Build generative AI applications inside Odokai that plug into workspaces, files, and agents. Early, working, and ready to show in a demo or pilot.

Software is becoming ephemeral, ubiquitous, and invisible. AI agents, the slow death of apps, and what I'm building with Odokai.

Introducing Odokai: a private AI platform for small to large UK businesses that need control over AI, usage, and workflow evolution.

Discover how enterprise employees can delegate daily tasks to AI agents, boosting productivity, reducing burnout, and freeing up time for strategic work. Learn practical implementation strategies.

Odokai helps UK businesses turn goals into outcomes with private AI control, governed usage, and operational automation.

From small businesses to enterprises, Odokai helps UK teams turn instructions into governed, real-world outcomes with private AI infrastructure.

Agentic AI tools are powerful but often complex. Odokai makes orchestration, workflows, and intelligent automation accessible to everyone.