AI on your own machines
When data cannot leave the building, we run the models locally instead.
Plenty of organisations have work AI could do, but data that cannot be sent to a third party in the cloud. Patient records, client files, sensitive registers, contract material.
That does not have to mean no to AI. It means the model has to come to the data rather than the other way round. We have built it and have it running, so this is not an opinion — it is an example.
What can run locally
Not everything sensibly can. We say what can, what hardware it needs, and where the line sits — before you invest in equipment.
Requirements and constraints
We start from what your legal team and your sector require, and build from there. Not the other way round.
A trial on your own material
We run a test on your real data, so you can judge the quality before the decision is made.
Models that run locally
Language and speech models installed on your own machines or servers. No data leaves the device at any point.
Dictation and transcription
Speech to text without the audio going anywhere. Including for languages the large providers handle poorly.
Running without an internet connection
Solutions that work on a machine with no network access, for the places where that is a requirement rather than a detail.
We have it running, not just on the drawing board
RepeatWhisper is our own dictation tool, where speech becomes text entirely on the machine itself. It is built for the languages the large providers handle poorly, and the audio never leaves the device. When someone asks whether local AI works in practice, we can show it rather than explain it.
Read about RepeatWhisperHow we deliver
The same approach whatever the area. You can stop after each phase, and you know what the next step costs before agreeing to it.
We map it out first
Ninety minutes going through your processes to work out what can genuinely be automated. You get a clear picture of what is possible and what it would cost.
We build on a fixed price
One- to two-week sprints, each ending in a demo. You know the cost before we start, and you own the code afterwards. No lock-in, no long contract.
We keep it running
Service plans covering maintenance, support and extensions as the business changes.
Is this the right place to start?
If two or more of these sound familiar, there is usually something to gain. If none do, we would rather say so now than after a proposal.
- Data must not leave the country or the organisation
- You work with personal or health information
- Your legal team has said no to a cloud solution
- You do not want to depend on a provider's pricing and terms
Questions we get asked
Are local models as good?+
Not at everything. On transcription, classification, summarising and extraction they are close. On tasks needing broad knowledge and long reasoning the large cloud models are still ahead. We say up front which category your task falls into.
What hardware does it need?+
Less than people expect. A lot of it runs fine on an ordinary modern Mac or PC. Bigger jobs need a server with a decent graphics card. We work that out as part of the assessment.
Who updates the model?+
We do, if you have a service plan. Otherwise you can, but it means somebody has to keep an eye on better models arriving — and they often do.