Which learners are genuinely ready to start
Enrolment evidence for around 450 learners a year, where funding and quality rules stay with qualified staff and the system only shows gaps.
A training provider turning over around one and a half million pounds, supporting roughly 450 learners a year.
A delivered Orchyn engagement. The client is anonymised at their request. The company and the people are not named. The operation, the workflow, the controls and the outcome are as delivered.
The commercial problem
Enrolment forms, identification, employer confirmations and learner evidence sit across email, a learning platform and shared folders. A coordinator chases the same missing items repeatedly.
The head of learner services has limited visibility of who is genuinely ready to start or at risk of missing a deadline. Quality checks get reconstructed later rather than captured as they happen.
How the work runs before
- A learner or employer submits documents by email or through the platform.
- A coordinator downloads and renames the files.
- They check them against the programme checklist by hand.
- Missing evidence triggers an email chase and a spreadsheet reminder.
- Complete files go to the head of learner services for a readiness review.
- The platform is updated manually and quality evidence is assembled after the fact.
What makes it hard to automate safely
- The model must not assess competence, make an admissions decision or issue a qualification.
- Personal information is minimised and handled under a clear retention and access policy.
- Funding and quality rules stay owned by qualified staff. The system shows gaps and cannot certify compliance.
- Every proposed record needs source evidence and a human review path.
What we change first
Onboarding was rebuilt around one learner evidence checklist with explicit ownership, deadline rules and a visible review queue. The checklist is the control. The model only populates it.
The redesigned workflow
- Approved learner evidence arrives into a secure intake.
- The model classifies the document and extracts the agreed fields, linking the source file.
- The workflow compares evidence against the programme checklist and marks each item present, missing or unclear.
- A readiness view shows which learners are complete, which are blocked and who owns the next action.
- The coordinator sends drafted chases. The head of learner services confirms readiness to start.
Where the model stops
The model classifies documents and populates a checklist. It does not assess a learner, approve an admission or confirm compliance. An unclear document is marked unclear, never assumed complete.
What we learned
The checklist did more than the model. Making readiness visible changed how the team worked. The extraction simply removed the typing. It is worth being honest that the cheaper intervention delivered most of the benefit.