On your own hardware
Runs on servers and devices you operate, including disconnected networks. Nothing is sent to us.
Solutions · Health and care
Patient data that never leaves the building.
Health and care teams handle huge volumes of referrals, messages and forms. Myles sorts and checks them on hardware you already operate, so patient data stays where it is, and anything uncertain goes to a clinician or administrator.
The problem
Data protection comes first
Patient information must stay within the organisation, and many teams cannot use cloud AI at all.
Administrative load is rising
Referral letters, patient messages and forms all need reading, sorting and routing before anyone can act.
Errors have consequences
An automated decision in a care setting needs a known error rate and a clear route to a person.
Where Myles sits
Why Myles
People decide the rest
Only decisions that clear a certified error rate are kept; everything else goes to a person.
A lineage you can check
A published record of the base model, data and licences, for your information governance team.
The evidence
91.8%correct on 1,400 held-out decisions, ahead of the medium tier of a leading hosted LLM family
2.7%error on the 73% of decisions kept at a 5% target, on unseen test data
£4per million decisions on one CPU thread, against £1,880 for that hosted tier
Myles-400M, our own measurement, September 2026. Method and all results