Education AI agent · State authorisation
State Authorisation & Licensure AI Agent
Keep the determination register a chief executive certifies from — which programmes lead to licensure in which states, the evidence under each row, the disclosure clocks — and never determine, disclose or certify.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- The CEO signs
- Never certifies
- After launch
- Agent Care
What this agent does
Builds the register, never the determination
In
- 01
A programme is offered into a state, and the register opens a row keyed on programme, credential level and CIP.
- 02
A determination changes, and the row keeps the old one, its date and the officer who made it.
Reason
- 03
A programme rests on institutional interpretation alone, and the row is flagged as such for the certifying officer.
- 04
A determination turns adverse, and a 14-calendar-day clock starts under 34 CFR 668.43.
- 05
A student moves state, and the row keyed on location at initial enrolment is re-opened rather than left.
Decide
- 06
A programme needs programmatic accreditation in a state, and that evidence is held apart from the licensure row.
- 07
A published state list omits the does-not-meet half, and the gap is raised in the pack.
Out
- 08
A certification year opens, and the pack is assembled clause by clause for the chief executive to read.
- 09
Execute write actions only inside the approval boundaries agreed during implementation.
Product statement
The agent maintains the register and the clocks; the chief executive personally certifies, and is the only person authorised to sign the funding application.
Example workflow
One programme, evidence to certification
- 1Programme in a stateCatalogue entry, enrolment location record, state board rule or a curriculum change
- 2Evidence gatheredEach state rule read, the source it came from, its date and the person who recorded it
- 3Register row assembledDetermination status per state, its provenance, staleness, open clocks and confidence
- 4Controls appliedProvenance checks, staleness checks, does-not-meet completeness and confidence threshold
No human action required
Stages 1 to 4 run unaided, and none of them determines that a programme meets a state requirement — the agent is assembling, and the officer lane opens at the confidence gate.
Goes to the compliance officer.
Adds a counsel read first.
Officer review
The row is held with its evidence, the provenance of each rule and the confidence.
Review · Correct · Send to counsel
Reviewed — the pack goes to the chief executive
- 6Catalogue and disclosure pages updatedOnly where write access and approval policy allow it
- 7Outcome evaluatedProvenance completeness, row corrections, notice timeliness and findings at recertification
Corrections
Every officer correction is counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Determining that a programme satisfies a state requirement for licensure.
- Making any disclosure to a student, prospective or enrolled.
- Certifying, or signing the participation agreement or the funding application.
- Deciding the state a student is located in under the institution policy.
Automation boundary
Agent acts unaided
- ✓Hold a row per state per programme with its rule cited.
- ✓Run the clocks from the determination and show the days left.
- ✓Flag which rows rest on institutional interpretation alone.
- ✓Show what the pack is missing, and hold it for the named officer.
Wrong after signing surfaces in a graduate who cannot sit the exam.
- Publishing the state list, or changing what the catalogue says.
- Records requests and registrar certifications.
- Accreditor notification and approval — our substantive change agent.
- Changes to clock, provenance or certification-control configuration.
Example output
One state row, annotated
Everything the agent assembles is attached to the source it came from.
Determination-register output · single programme
Illustrative example- Programme
- Nursing, pre-licensure
- Evidence line
- Clinical hours in the state board rule sit above the hours the current curriculum awards
- Days to notice
- Day 3 of 14
- Rule applied
- 34 CFR 668.43
- Confidence
- 86%
- Determination status
- Adverse — notice not yet sent
As received
Taken from the state board rule and the catalogue as published — nothing on this side is the agent's.
Source used
- State board rule text
- Catalogue curriculum
- Enrolment location record
Why this flag
The rule and the curriculum disagree on hours; the officer decides what that means.
What the score decides
Below the configured threshold the row picks up a counsel read before the officer sees it.
Action
ReviewCorrectSend to counselValue
Where AI adds value
The same four claims, placed at the point in the workflow where each one applies.
Where the value lands
Value 01 – 04
Every programme row
From the catalogue and the state rules
Assemble against the rule
Draw on the catalogue, the state board rules on file and the enrolment location records. Under 34 CFR 668.14 the institution must determine, state by state and programme by programme, that a programme satisfies that state's educational requirements for licensure.
The chief executive signs it
Routine row assembly and clock arithmetic arrive already done.
Point people at unvouched rows
Most determinations in this sector rest on institutional interpretation and few are board-confirmed — yet the signature block names the chief executive under a criminal-penalty acknowledgement, and only that officer may sign the funding application. The agent flags those rows; it never fills one in.
Build an evidence trail
The programme, the state rule it was read against and the officer who certified stay on the file.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
Student systems
- Banner
- Colleague
- Workday Student
- Jenzabar
Catalogue and curriculum
- Acalog
- CourseLeaf CIM
- Coursedog
- Kuali
State authorisation records
- SARA portal filings
- State board rule files
Agent
State authorisation and licensure
- Reads the rules
- Assembles the register
- Holds for the officer
Web and disclosure pages
- Drupal
- WordPress
- Programme disclosure pages
Observability & evaluation
- OpenTelemetry
- Langfuse
- Supported monitoring/evaluation sources
Integration availability depends on the client's existing systems and API access.
Agent controls
Six layers between the model and the certification
Six checks, and each one assumes the last held. What defeats them all is in the map below.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether a row may stand
- L3
- Puts the row in the officer's hands
- L4 – L5
- Keep the programme and the rule behind it
- L6
- Narrows to determination assembly when signals degrade
How Nestack evaluates it
Evaluate the whole register — not only the row on the screen.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the pack the chief executive reads
- E1Final-output evaluationDid each row carry the state rule it was read against?
- E2Step-level evaluationDid the agent use the right catalogue version, state rule and location policy?
- E3Tool evaluationDid it read the correct programme and write the correct row?
- E4Confidence calibrationDo low-confidence rows actually attract more officer corrections?
- E5Slice evaluationHow far apart do provenance rates sit across specific programme types?
- E6Business outcomeHow many rows were corrected, and how many surfaced only at recertification?
Floor — the certification the chief executive answers for
Failure modes
Where each failure originates in the agent
Seven failure modes, set at the lifecycle stage where each one begins.
Agent lifecycle
Direction of processing →
01 · Retrieval
1 mode
- HO-03
Superseded state rule
A row is built from a board rule that has been revised.
Stage gathers
Catalogue, state rules and location records
02 · Reasoning
2 modes
- HO-04
Programme name keying
A certificate inherits the degree it was spun out of.
- HO-06
Interpretation unflagged
A row resting on one reading is presented as evidenced.
Stage proposes
Determination status, provenance and confidence
03 · Tool / write
2 modes
- HO-02
Adverse row runs quiet
A newly adverse row starts no clock at all.
- HO-05
Half the list published
The does-not-meet states are dropped from the list.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- HO-01
No determination on file
A Title IV-eligible programme has a state with no row.
Stage returns
The register the chief executive certifies from
05 · Change / Version
1 mode
- HO-07
Curriculum change unmapped
A course substitution never reaches the state rows.
Stage tracks
Model, prompt, clock rules and state-rule config
- Sev-1 · a determination made without a person
- Sev-2 · a notice is missed or a list is wrong
- Sev-3 · a rule goes stale, row goes to review
Affected slices
One programme type can carry most of the stale rows
A catalogue-level determination-currency figure can read sound while a few states and a few programme types hold most of the stale rows. Nestack reports the staleness rate by slice, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Pre-licensure nursing states | 7.3% | 3.6× | Review | |
| Teacher preparation states | 5.1% | 2.5× | Review | |
| Behavioural health states | 3.6% | 1.8× | Watch | |
| Home-state non-licensure | 2.4% | 1.2× | Normal |
Bar: staleness-rate lift vs. the home-state non-licensure baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
A missed notice ends as a test
A cycle is done when the missed disclosure has become a case the next release must pass. That suite is what the next determination made is measured against.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Staleness rate rises in one programme type.
- 02
Diagnose
The programme that qualified a graduate in thirty states and not the one she lived in is read back through the register until the cause narrows to one.
- 03
Improve
Stamp the change; the programmes behind it are filed against that number.
- 04
Verify
Each touched programme case is run again, and one red stops the release.
- 05
Learn
It stays a standing test, and the state rules move with it.
Learn → Detect
The return edge. The next determination is checked against a suite one case longer.
Typical build scope
Twelve workstreams across six weeks
The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, sources, determination workflow, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Certification workflow discovery and the boundary.Week 1
- 02Catalogue and student-system assessment.Week 1, Week 2
- 03Clock, provenance and state-rule mapping per state.Week 1, Week 2
- 04Catalogue and location-record ingestion.Week 2
- 05Register assembly and rule binding.Week 2, Week 3
- 06Confidence scoring and flag routing.Week 3
- 07Officer review workflow.Week 3, Week 4
- 08Catalogue and disclosure-page integration.Week 5
- 09Determination and disclosure cases.Week 4, Week 5
- 10Guardrails and signing controls.Week 4
- 11Programme-trail instrumentation.Week 5, Week 6
- 12Deployment, documentation and Agent Care handover.Week 6
12 workstreams · 6 weeks · bar shows the weeks a workstream is active — several run in parallel
Final scope and sequence confirmed in discovery
Engagement tiers
What each tier includes
Rows are the capabilities named in each tier's scope. Higher tiers include everything below them.
| Capability✓ in scope · — not at this tier | PilotOne catalogue, one state set | ProductionProduction catalogue and student systems | AdvancedMultiple campuses / catalogues |
|---|---|---|---|
| Introduced at Pilot | |||
| Rows and clocks against your rules | ✓ | ✓ | ✓ |
| Officer review | ✓ | ✓ | ✓ |
| Determination-currency baseline | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Reporting by state | — | ✓ | ✓ |
| Review workflow in your systems | — | ✓ | ✓ |
| Approved page write-back | — | ✓ | ✓ |
| Catalogue-and-programme integration | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Complex multi-state rule sets | — | — | ✓ |
| Multi-stage officer review | — | — | ✓ |
| High programme count | — | — | ✓ |
| Multi-state disclosure controls | — | — | ✓ |
| Build price | From $5,000 | From $8,000 | Custom quote |
- Final build price
- Confirmed after discovery based on integrations, workflow complexity, programme count, review controls and deployment requirements.
- Separate from build
- Build pricing is separate from recurring Agent Care, which covers managed monitoring, evaluations, incidents and verified improvements after launch.
What we need from you
What you bring, and what we build with it
Each input maps to a piece of build scope and a week in the delivery timeline.
You bring
We build with it
01Your programme inventory and catalogue structure
→Catalogue and location-record ingestion, and rule binding
Week 102Representative determinations from a past year
→Determination-currency baseline, provenance capture and row keying
Week 203Your state rule files and where they are kept
→Clock, provenance and state-rule mapping per state
Week 104Access to relevant APIs, feeds or exports
→Catalogue, student-system and web review, then integration setup
Week 205Determinations you would not want tested
→Notice cases and the standing evaluation suite
Week 406What no certification may take on trust
→Confidence scoring, flag routing, guardrails and signing controls
Week 307The named officer, and the chief executive who signs
→Officer review workflow, then pilot and production validation
Weeks 5–6
Nothing else is required
Deployment, documentation and Agent Care handover are ours.
Delivery timeline
Four phases across six weeks
Weeks genuinely spent set each one of these widths, so the fifth week legitimately holds a pair.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Certification workflow discovery, rule mapping and the boundary
W2Catalogue integration and the determination-currency baseline
W3Register assembly, clock arithmetic and review controls
W4Evaluation suite, notice cases and failure-mode testing
W5Disclosure-page integration, a pilot state set and targeted corrections
W6One certification year run under the chief executive, then Agent Care handover
- Reading the band
- A bar covers only the weeks its own work is named in. The fifth-week doubling is real, not padding.
- At the end of W6
- When the certification pack validates, Agent Care takes the agent on.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Education AI agent
Build a licensure certification agent around the chief executive who signs.
One signature covers the whole instrument, and the hardest clause on it is that a programme meets the licensure requirements of each state your online students live in. Show us your programme inventory, your state rule files and who certifies.
- Typical delivery
- 4–6 weeks
- Deployment
- Your catalogue and student stack
- The CEO signs
- Never certifies
- After launch
- Agent Care available
Nestack Agents · State authorisation and licensure
AGT-ED-14 · Agent Care available after launch