Companies
We work with a select few companies per industry. A collaboration starts with historical data and a defined measure, before anything runs live.
Contact us
Thirty minutes to work out where to start.
The first conversation takes thirty minutes. We look at the environment and the data, and work out whether there is a learning task the AI fits. There is nothing to prepare.
Email usWho we work with
Three kinds of conversation start here.
Researchers
Our scientists come from physics, nonlinear dynamics, computer science and neuroscience. If you want to work on continual learning with us, write to us.
Investors
The thesis, the evidence and the company facts are on the investors page.
InvestorsHow a collaboration runs
-
1 · First conversation
Thirty minutes, nothing to prepare. We look at the environment and the data and work out whether there is a learning task the AI fits.
-
2 · Retrospective
With someone who knows what the values mean, we choose the learning task and run an instance on historical data, against a defined measure.
-
3 · Live
An instance runs on a live process, on the infrastructure where the data lives. This is where we need access to the operational data.
-
4 · Result
We measure the effect in the partner’s own operation and decide together how to continue.
What stays where it is
The data never leaves the environment it is collected in. What the instance learns stays next to that data, and the decisions stay with the software around it.
Get in touch
Worth mentioning in your email:
- Which environment and which measured values are involved, and how often does a new value arrive?
- Which decisions rest on values that were set once, and when were they last checked?