An AI engine that follows your method.
Aikila does a piece of work the way you would do it, every time, and shows how it got there.
[A real flow: compliance review, laid out on the canvas]
Most AI agents find their own path. Aikila follows yours.
For the work that has to be done a particular way, and be seen to have been: you write the method down once, as blocks on a canvas, and the engine follows it.
- Reasoning only
It never touches your systems.
It reads what it is handed and writes the result. No email, no browsing, no acting - that stays with the systems and people around it.
- Your method
The steps are yours, written down once.
No steering it through chat. The method lives as blocks on a canvas, and the next file is handled exactly like the last.
- On the record
It shows how it got there.
Every step, judgement and source ends up in a trace you can read - so the result can be checked, not just believed.
In the browser, from your stack, or under a product.
In the browser
Lay the method out on the canvas, hand it files, get the deliverables back with the full trace.
From your stack
Called over REST or MCP: the reasoning step inside the automations you already run.
Under a product
A foundation for stand-alone AI products, where the reasoning has to be trusted first.
What it looks like in practice.
Some examples of Aikila flows
Three bodies of text that all have to agree with each other.
What the organisation actually does, what its policies say it does, and what the law requires. Anna can describe her review precisely - she has done it for years. She just cannot get through it quickly, and it has to hold up in front of a regulator.
Closed for now, and by invitation.
Tell us what you would want to automate and we will send you a code. If you only want to look, the demo is open to anyone.