Grow the humans running your AI agents.
Adopting AI is a question about humans before it is a question about tools. This is a registered apprenticeship: somebody learns to build and run AI solutions on your real client work, with a mentor beside them, and moves on when they can show they can do the job rather than when the calendar says so.
You move on by what you can do, not by the calendar
The programme is registered with a national apprenticeship register, and an apprentice advances by showing they can do the thing, with a mentor signing it off, rather than by serving a set number of months. Classroom instruction and hours on the job are counted separately, because the registration requires it.
Enrol an apprentice, and they read nothing until they sign
Enrolling somebody puts them on a project under a mentor and a cohort in one go, and sends them a preview and confidentiality agreement to sign, in the same way a client signs one. The course material is ours and is licensed, not published, so nothing opens for them until they have read that agreement end to end and signed it by typing their name. There is no page to skip past: the lock is on our side.
- The curriculum version is pinned at enrolment; a later release never rewrites what somebody was already taught.
- An apprentice sees only what is theirs or shared with them.
The curriculum, and a book that can be made from it
An instructor reads the published curriculum as courses and modules, and every module says plainly whether its lesson is registered content, a draft still waiting on a human, or not written yet. Nothing is hidden to look finished. The same version can be composed into a book on request, built fresh each time from what is published. There is deliberately no download yet: how licensed material leaves the platform is a human's decision.
Hours on the record, by kind
The platform already keeps on-the-job hours and instruction hours apart, each one approved by somebody other than the human who logged it. That is what the apprenticeship reporting is built on: a human wrote the hours down, a reviewer approved them, and nothing is guessed at.
A mentor's queue, not a chat message
Short answers wait for a mentor to mark them, the solution sitting beside each one so grading is not a memory test, and competencies wait to be signed off against the evidence an apprentice submitted. A mentor sees only their own apprentices' queue, never the whole cohort's.
The record the sponsor reports
For every apprentice, the platform keeps the competencies signed off and by whom, hours against the registered totals, the wage step reached and completion, and it exports that record on request, as a read that is itself recorded.
Ask about hosting an apprentice, or enrolling your first one.
Also in AI management system
- AI agents and their workGive an AI agent a job, then read back every step of it.Read the page
- SandboxA machine of its own per session, and your keys stay out of it.Read the page
- Command lineThe terminal your engineers already use, on your rules.Read the page
- MCP serverConnect any AI agent, and the rules come with it.Read the page
- Client SDK and APIOne typed client for the whole platform, given to your team.Read the page