NEET processes — plain English
Synced from Asana (Technology & Data, Process Design, Roles, AI Specialist Design — 20 Jul 2026). Door-opener view; Asana remains the living design pack.
How to use this with Peter / partners
Walk the stages, open a demo Need, then show Specialists (including the AI catalogue). This is the operating model made visible — not a live caseload. Formal SOPs and approvals still finish in Asana.
App status spine (Prototype 1)
Plus continuous improvement (Leave It Better) at any stage.
Full OS lifecycle (Asana) ↔ this prototype
Asana defines a richer lifecycle for the funded product. Prototype 1 folds it into fewer board statuses so demos stay clear.
| # | Asana / full OS stage | Shown in app as |
|---|---|---|
| 1 | Need Identified | Received |
| 2 | Need Acknowledged | Received |
| 3 | Need Understood | Triaged |
| 4 | Need Assessed & Triaged | Triaged / Assessed |
| 5 | Need Approved | Assessed |
| 6 | Planning | Matched |
| 7 | Specialist Matching | Matched |
| 8 | Intervention Delivery | Intervention active |
| 9 | Testing & Validation | Outcome pending |
| 10 | Need Fulfilled | Outcome confirmed |
| 11 | Benefits Confirmed | Outcome confirmed |
| 12 | Sustainment | Sustainment |
| 13 | Closure | Closed |
| 14–19 | Leave It Better (review → Blueprint) | Improvements |
Operating rules (non-negotiables)
Assessments, matches, interventions, evidence and outcomes hang off the Need.
One operational record — not twelve partner spreadsheets.
Ask what capability is required first. Then choose who or what delivers it.
Unsafe → escalate. Jobs and training wait.
A session or referral is activity. An outcome is a confirmed change with evidence.
Vacancies and trials sit on the same board as young person Needs.
Process stages (door-opener view)
1. Intake — “What do you need?”
Every story starts with the Need. Capture what was asked for, who it concerns, and the channel — once.
- Record title + description in the person’s / referrer’s own words
- Note source, subject, desired outcome if known
- Flag immediate risk indicators on first contact
- System assigns a stable Need ID / reference
- Jumping to a course before naming the Need
- Orphan notes not attached to a Need
2. Triage — universal front door
Every Need passes a consistent triage: presenting vs underlying Need, urgency, risk, eligibility route, owner. AI may recommend; a qualified specialist decides.
- Immediate risk screen (safeguarding wins)
- Set priority, risk, classification, next route
- Assign an owner — nothing unowned
- Acknowledge the person where appropriate
- High-risk with no owner
- Treating every Need as “get a job”
3. Assessment — proportionate understanding
Light-touch when clear; deeper when risk, complexity or barriers require it. Ends with capability required and intended outcomes.
- Strengths, aspirations, skills, barriers
- Existing support and accessibility needs
- Define capability required before naming a provider
- Over-assessing simple Needs
- Assessing forever instead of acting
4. Specialist matching — capability first
Match the capability, then the specialist. Specialists include people, employers, colleges, AI, tools, automation and partners.
- Search directory by capability
- Score fit; allow multi-specialist packages
- Human approves material matches
- Employer vacancies are Needs on the same board
- Referral to whatever has spare places
- Hiding that AI proposed a match
5. Intervention & delivery
Plan intended outcome, activities, milestones and evidence. Log progress. Stalls are operational data.
- Link intervention to Need + specialist
- Name intended outcome before start
- Remove practical blockers early (travel, childcare)
- Log missed activity honestly
- Counting appointments as success
- Quiet drop-off with no note
6. Outcome confirmation
Activity completed ≠ outcome achieved. Authorised person confirms with evidence. Alternatives can still be positive.
- Separate activity, intervention completion, outcome decision
- Attach evidence for significant claims
- If waiting on employer, leave pending and chase
- Closing as success because they attended
- No evidence for big claims
7. Sustainment
Outcomes can fail at week 4 or 13. Plan check-ins so starts stick.
- Schedule sustainment reviews
- Record still-on-track or not
- Re-open or raise a new Need if things fall apart
- “Job done” on day one of a placement
8. Leave It Better — close & improve
Close with a reason. Capture what worked / failed. Approved learning updates the Blueprint — not just one case.
- Close only when criteria met (Asana: no silent zombies)
- Log improvements local or Blueprint-wide
- Pass learning forward
- Lessons stuck in chat or someone’s head
Forms & records (from Asana framework)
Core principle: a form captures information; a record preserves evidence, decisions and history.
Capture “What do you need?” once — subject, source, description, desired outcome.
Presenting vs underlying Need, risk, urgency, route, owner. AI may recommend; human decides.
Proportionate strengths, barriers, aspirations, capability required.
Capability fit, scores, selected specialist(s), approval where required.
Intended outcome, activities, milestones, evidence required, review dates.
What happened, missed activity, blockers — stalls are data.
Proof for significant outcomes (offers, certificates, attendance).
Activity ≠ outcome. Authorised person confirms with evidence.
Is the outcome still true at 4 / 6 / 13 weeks?
Close cleanly; capture learning for the Blueprint.
When AI recommends, record what it said and whether a human accepted or overrode it.
Who does what (RACI roles from Asana)
One Accountable owner per activity. AI can be Responsible for analysis — never Accountable for professional judgement alone.
| Code | Role | In practice |
|---|---|---|
| YP | Young person | Consulted throughout — with them, not to them |
| TS | Triage specialist | Front door ownership and risk screen |
| CS | Case specialist (youth support) | Day-to-day Need ownership through delivery |
| SM | Service manager | Approvals, exceptions, outcome confirmation |
| SL | Safeguarding lead | Accountable for safeguarding responses |
| EM | Employer engagement | Employer Needs, vacancies, trials |
| RP | Referral partner | Hand-off once; stay informed |
| PO | Process owner | Process health and Blueprint improvements |
| AI | AI specialist | Responsible for analysis/recommend — never Accountable alone |
| AUTO | Automation | Executes approved workflows; never Accountable |
AI Specialist rules (Asana AI design)
Assign work by the capability required — not by “should a human or AI do this?” as the first question.
AI may assist, analyse, recommend, draft, organise or alert. Material decisions stay with an authorised person.
Users should be able to tell when content or recommendations came from an AI Specialist.
AI recommendations must not quietly become final decisions without review.
AI only receives the data needed for its approved purpose.
Every AI Specialist must know when to stop and transfer to another specialist — especially safeguarding.
AI catalogue (18) — also on Specialists
- AI Welcome and Navigation
- AI Intake
- AI Triage
- AI Assessment Support
- AI Capability Identification
- AI Matching Specialist
- AI Planning
- AI Communication
- AI Meeting and Session Support
- AI Evidence
- AI Outcome Analysis
- AI Sustainment Monitoring
- AI Quality Assurance
- AI Data Quality
- AI Reporting and Insight
- AI Continuous Improvement
- AI Knowledge
- AI Safeguarding Support
Open specialist directory →Young person + AI interaction model & stakeholder journeys →
Young person experience goal: “I was listened to, I understood what was happening, I remained involved in the decisions, I could speak to a person when I needed to, and the support helped fulfil my Need.”
Suggested demo walk
- Board view — Needs
- Safeguarding first — NEED-2026-0004
- Capability match — NEED-2026-0001 or 0008
- Stall + learning — NEED-2026-0006 → Improvements
- Outcome with evidence — NEED-2026-0013
- Employer Need — NEED-2026-0011
- AI catalogue — Specialists (filter kind AI)