PHASE 01
Co-design and approvals
Work with people with lived experience, educators, prison or probation staff and employers. Define one cohort, existing provision, recruitment, accessibility and operational constraints.
THE UNIVERSITY RESEARCH PROPOSITION
Test the pathway.
Challenge the claims.
We need an independent partner to help establish whether the intervention is useful, deliverable and worth investing in.
THE CENTRAL RESEARCH QUESTION
Does a clearer future
change participation today?
Can personalised learning, verified progression and an employer-linked resettlement pathway improve engagement and sustained employment, and what public value is generated per pound invested?
Study the whole intervention first. To isolate the contribution of AI, a later design could compare equivalent human-led support with and without the AI component.
PROPOSED DESIGN
Feasibility first.
Impact with the right design.
PHASE 02
A supervised feasibility test
Indicative 100 participants; final sample and timeline agreed with the university. Test usability, uptake, delivery fidelity, staff burden, safety and data completeness before claiming impact.
PHASE 03
An independent outcome study
Use random allocation if ethical and operationally feasible, or a defensible comparison design. Pre-specify outcomes, statistical power, follow-up, missing-data handling and limitations.
WHAT WE WOULD MEASURE
Outcomes that can be checked.
| Question | Measure | Evidence approach |
|---|---|---|
| Participation | Recruitment, retention, active use and completion | Programme logs plus participant and staff interviews |
| Learning and confidence | Verified qualifications, reading engagement and job-readiness measures | Provider verification and agreed validated measures where appropriate |
| Transition into work | Interviews, job starts, earnings where permitted, retention at 3 / 6 / 12 months | Employer or administrative verification where accessible; record verification gaps |
| Resettlement | Housing stability, appointments and successful referrals | Partner records and participant feedback with an agreed lawful data route |
| Delivery and economics | Full delivery cost, net staff time, cost per sustained job | Time sampling, cost ledger and comparison with existing support |
| Long-term justice outcomes | Proven reoffending, frequency and time to event where data allow | Pre-agreed official definitions, comparable cohorts and sufficient follow-up |
Engagement and job offers are intermediate outcomes. They are not substitutes for sustained employment or independently assessed reoffending outcomes.
RESEARCH GOVERNANCE
Earn trust through
how the study is run.
Approval before participant research
University ethics and relevant HMPPS National Research Committee approval are part of the proposed route. Operational support from a prison does not replace research approval.
HMPPS research guidanceParticipation without pressure
Research participation would be voluntary and separate from privileges, progression decisions and release. Agree accessible consent, withdrawal arrangements and safeguards for the custodial power imbalance.
Data and AI governance
Agree controllers, lawful bases, criminal-offence and any special-category data conditions, retention, role-based access, audit trails and a data protection impact assessment. Research consent does not by itself settle the lawful basis for every processing activity.
Test AI accuracy, bias and accessibility; use approved content and staff escalation. No automated release decisions, risk scoring or clinical diagnosis are proposed. Approval of security architecture and devices is a prerequisite to custodial deployment.
WHAT THE UNIVERSITY GETS
A substantive applied
research collaboration.
ACADEMIC CONTRIBUTION
A question worth investigating
An opportunity to study motivation, learning, continuity of care and employability at the intersection of criminology, education and responsible AI.
INDEPENDENCE
Publish what the evidence says
Agree publication rights, authorship, negative-result reporting and access to sufficient data before the study starts. Avoid commercial control over research conclusions.
KNOWLEDGE EXCHANGE
A route from research to practice
Co-develop fundable work packages and useful guidance for frontline services, with clear agreements on background IP, project IP and licensing.
AN INVITATION TO UNIVERSITIES
Help turn a promising idea
into credible evidence.
We are seeking a university research lead, an operational justice partner and employers willing to help design the first test.