Main site

Education AI agent · Family contact

Parent-Communication AI Agent

Send routine school messages to the contacts a rights check has cleared, hold anything legally operative for a person, and treat a custody or safety restriction as a limit the agent cannot cross.

Typical delivery
4–6 weeks
Deployment
Your stack
Person releases
Rights-checked
After launch
Agent Care

What this agent does

Sends the message, not the notice

In

  1. 01

    Ingest contacts, enrollment status, language preferences and restriction flags from supported SIS.

  2. 02

    Normalise households and contact roles, and carry each recipient forward with the record that authorises them.

Reason

  1. 03

    Re-check entitlement before every send: rights pass to the student at majority or on entering postsecondary study.

  2. 04

    Check each recipient against custody orders, protective orders and safety plans, and fail closed on a stale flag.

  3. 05

    Apply the district's message categories, consent records and opt-outs, and honour a revocation across all of them.

Decide

  1. 06

    Hold any notice that carries a legal consequence — discipline, special education, a deadline — for a person to release.

  2. 07

    Route a family's language need to a qualified human translator.

Out

  1. 08

    Retain the rights check, the recipient list, the release and the delivery record against the message.

  2. 09

    Execute write actions only inside the approval boundaries agreed during implementation.

Product statement

The agent sends routine notices to cleared contacts; a person decides who is entitled to a record, and what an operative notice says.

Example workflow

One message, request to delivery

  1. 1Message requestedAttendance event, calendar change, staff request or special-education workflow
  2. 2Entitlement assembledHousehold contacts, entitlement status, restriction flags and language preference, each with its source
  3. 3Message preparedCategory, recipient list, language route and confidence
  4. 4Rights checks runEntitlement re-check, restriction and safety-plan check, category and consent rules, and confidence threshold

No human action required

Nothing is sent in the first four stages, and no recipient is cleared by the agent alone — the release lane opens at the confidence gate.

5DecisionSplits on the rights check
Recipient confirmed

Goes to the named staff member to release.

Recipient in doubt

Adds a records-office check first.

Staff release

The message is held with its recipient list, the checks it passed and the confidence.

Release · Correct · Send to the records office

Released — delivery recorded

  1. 6Notification systems updatedOnly where write access and contact policy allow it
  2. 7Delivery evaluatedEntitlement misses, restriction breaches, opt-out compliance and errors found after delivery

Recalls

Every staff correction is counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required

8 items

  • Deciding whether a parent may receive an adult student's record.
    • Determining a student's tax-dependency status.
    • Resolving a custody, protective-order or contact-restriction question.
    • Making a health-or-safety emergency determination.

  • Automation boundary

    Agent acts unaided

    • ✓Send closure, delay and schedule-change notices to cleared contacts.
    • ✓Send unexcused-absence notifications inside.
    • ✓Confirm event responses and conference bookings.
    • ✓Answer published calendar, bell-schedule and transport questions.

    Any send happens inside the boundaries agreed at implementation, never ahead of the rights check.

    • Authoring or approving special-education prior written notice.
    • Approving the translation of a legally operative notice.
    • Making a mandated child-abuse or neglect report.
  • Communicating a disciplinary outcome or its appeal rights.

Example output

One outbound message, annotated

Everything the agent prepares is attached to the record that authorised it.

Communication output · single message

Illustrative example
Message
Early dismissal
What it says
Your child's day ends early on Thursday and the bus runs to the altered timetable
Recipients cleared
One guardian, one held
Source of record
SIS contact record, current
Confidence
93%
Language route
Qualified translator queued

As received

Taken from the contact record and the restriction flags on file.

Checks used

  • Entitlement status on file
  • Custody and safety flags
  • Language preference on file

Why this recipient

One guardian is cleared and one is held by an order and a person still decides.

What the score decides

Below the configured threshold the message picks up a records-office check before anyone.

Action

ReleaseCorrectSend to the records office

Value

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 outbound message

From the contact record

03Preparation

Prepare from the record

Getting the recipient wrong is not a paperwork problem — it is a child's schedule or location reaching an adult a court has excluded.

01Approved path

Reach the guardian who matters

Routine closures, schedule changes and absence notices go out already checked.

02Human review

Send staff where the risk is

Restricted households, transferred rights and language needs are marked, so staff time goes to the sends that can do harm.

04

Build an evidence trail

Each message keeps the rights check behind it and who released it.

Integrations

Typical integrations

Five system groups connect to the same agent. Which of them are in scope is decided in discovery.

  • Notification platforms

    • ParentSquare
    • Remind
    • Blackboard Connect
    • SchoolMessenger

  • Student information systems

    • PowerSchool SIS
    • Infinite Campus

  • Classroom communication

    • ClassDojo
    • Seesaw

Agent

Parent communication

  • Reads the record
  • Prepares the message
  • Holds for release

  • Special education

    • Frontline Special Education
    • PowerSchool Special Programs

  • 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 send

Each wraps the last. The map below names what each one lets through.

L6 · Outermost — last line of defence

Inward → L1 · closest to the model

L6Rollback / safe modeHold every send and queue it for a person when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt, category-rule and contact-configuration changes.Track
L4TraceabilityRecord the checks, the recipient list, the release and the delivery time.Record
L3Staff releaseHolds the message for a named person; it governs who releases, not whether a delivered message can be taken back.Gate
L2Policy guardrailsTest each send against entitlement, restriction flags, category and consent rules; a failure returns the message.Restrict
L1Confidence thresholdsRoute low-confidence recipient lists to a records-office check before release.Require review
Model coreMessage prepared — category, recipient list, language route and confidence
L1 – L2
Test whether a message may stand
L3
Puts the release in a person's hands
L4 – L5
Preserve who was told what, and on what basis
L6
Holds sends and queues them for a person

How Nestack evaluates it

Evaluate the whole send path — not only the message that lands.

Coverage runs the whole depth of the workflow, and every layer is cut by slice.

Surface — the message the family receives

  1. E1Final-output evaluationDid every recipient clear the entitlement and restriction checks?
  2. E2Step-level evaluationDid the agent use the right contact record, flags and language preference?
  3. E3Tool evaluationDid it read the correct household and write to the correct delivery log?
  4. E4Confidence calibrationDo low-confidence recipient lists actually attract more staff corrections?
  5. E5Slice evaluationHow far apart do error rates sit across specific household types?
  6. E6Business outcomeHow many messages were corrected, and how many were caught only after delivery?

Floor — the message the family receives

Failure modes

Where each failure originates in the agent

Seven modes across the delivery lifecycle.

Agent lifecycle

Direction of processing →

  1. 01 · Retrieval

    1 mode

    • GD-03

      Cached contact list

      A list that predates a restriction flag includes the held parent.

    Stage gathers

    Contacts, entitlement status, with the source each came from

  2. 02 · Reasoning

    2 modes

    • GD-04

      Transfer of rights missed

      A dual-enrolled minor is treated as covered by parental rights.

    • GD-06

      Category misread as urgent

      An engagement campaign goes out as attendance messaging.

    Stage proposes

    The category, the recipient list and confidence

  3. 03 · Tool / write

    2 modes

    • GD-02

      Machine-translated notice sent

      A suspension notice goes out translated, its deadline wrong.

    • GD-05

      Message reads as formal notice

      A meeting summary starts a safeguards clock nobody authorised.

    Stage writes

    Only where write access and approval policy allow it

  4. 04 · Output

    1 mode

    • GD-01

      Reply reads as closure

      A parent's safety disclosure is answered, and no person is paged.

    Stage returns

    The message the family reads and acts on

  5. 05 · Change / Version

    1 mode

    • GD-07

      Opt-out preferences reset

      A platform upgrade resumes messaging to families who revoked.

    Stage tracks

    Model, prompt, category rules and contact config

  • Sev-1 · a held adult receives a child's
  • Sev-2 · an operative notice goes out unverified
  • Sev-3 · contact data degrades
See the complete Education failure-mode catalogue

Affected slices

A low average over a very uneven set

The baseline is the error rate across every message the district sends, and it is low. Three cohorts sit several times above it Nestack reports it by slice rather than in aggregate..

Slice performance — reported separately, not only in aggregate

Illustrative example
SliceFailure rateLiftLift vs. thresholdStatus
Households with a restriction on file7.0%3.9×Review
Notices to families with a language need6.1%3.4×Review
Students whose rights have transferred2.7%1.5×Watch
Routine notices to cleared contacts0.9%0.5×Normal

Bar: error-rate lift vs. the all-message baseline · scale 0–4.0× · tick marks the 2.0× review threshold

2 of 4 slices over threshold

Evidence-linked improvement

A coordinator cannot close a cycle

A cycle is finished when the failure sits in the suite as a case the next release must pass.

Improvement cycle · five stages

Switchback — the path turns at Improve and returns at Learn

  1. 01

    Detect

    Error rate rises in one household cohort.

  2. 02

    Diagnose

    The sends, the contact history and the flag version are read together until one of them explains it.

  3. 03

    Improve

    The correction carries its version and the messages behind it.

  4. 04

    Verify

    The affected cases are re-run before anything ships.

  5. 05

    Learn

    It lands in the permanent suite and in the contact rules.

  6. Learn → Detect

    The return edge. Every later detection runs against the longer suite.

Typical build scope

Twelve workstreams across six weeks

The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, contacts, message workflow, evaluation, integration, then production validation and handover.

Workstream

Week 1Week 2Week 3Week 4Week 5Week 6

  1. 01Communication workflow discovery and scope.Week 1
  2. 02SIS and notification-platform assessment.Week 1, Week 2
  3. 03Entitlement, restriction and message-category mapping.Week 1, Week 2
  4. 04Contact ingestion and household resolution.Week 2
  5. 05Message assembly and rights-check.Week 2, Week 3
  6. 06Confidence scoring and release routing.Week 3
  7. 07Staff release workflow.Week 3, Week 4
  8. 08Notification- and SIS-platform integration.Week 5
  9. 09Delivery regression cases.Week 4, Week 5
  10. 10Guardrails and rights controls.Week 4
  11. 11Delivery-trail instrumentation.Week 5, Week 6
  12. 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 tierPilotOne school, one message setProductionProduction notification systemsAdvancedMultiple schools / districts
Introduced at Pilot
Messages to your contact rules✓✓✓
Staff release✓✓✓
Delivery baseline✓✓✓
Introduced at Production
Reporting by household—✓✓
Release workflow in your systems—✓✓
Approved write-back—✓✓
Notification-platform integration—✓✓
Introduced at Advanced
Multi-district and multi-state rules——✓
Multi-stage release approvals——✓
High message volume——✓
Enterprise contact controls——✓
Build priceFrom $5,000From $8,000Custom quote
Final build price
Confirmed after discovery based on integrations, workflow complexity, message volume, release 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.

  1. 01Your contact records and where restrictions live

    Entitlement and restriction mapping

    Week 1

  2. 02Representative sends from a past term

    Delivery baseline, household normalisation and rights-check binding

    Week 2

  3. 03Your message categories and consent records

    Entitlement, restriction and message-category mapping

    Week 1

  4. 04Access to relevant APIs, feeds or exports

    SIS, notification and special-education assessment, then integration setup

    Week 2

  5. 05Messages that reached the wrong adult

    Delivery cases and failure-mode testing

    Week 4

  6. 06Which messages may never send without a rights check

    Confidence scoring, release routing, guardrails and rights controls

    Week 3

  7. 07Named staff to release held messages

    Staff release 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

Phases sit where the work is, which is why week 5 holds evaluation and pilot at the same time.

Phase

W1

W2

W3

W4

W5

W6

  1. DiscoveryW1
  2. BuildW2 – W3
  3. EvaluateW4 – W5
  4. Pilot & LaunchW5 – W6

Week focus

W1Communication workflow discovery, contact mapping and the automation boundary

W2Contact integration and the delivery baseline

W3Message workflow, confidence logic and rights controls

W4Evaluation suite, restriction checks and failure-mode testing

W5Platform integration, a pilot term and targeted corrections

W6A term of family communication under the school, then handover

Reading the band
Bands cover the weeks their work is named in and no more. The doubled fifth week is real, not padding.
At the end of W6
Everything verified on live sends, with Agent Care carrying it forward.
Duration
Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Education AI agent

Build a communication agent around your district's contact rules.

Show us your notification platform, your contact records and where custody restrictions are held. If your protective-order flags sit somewhere the messaging platform never reads, that is the first thing to fix; then we map the routing, set the automation boundary and name what stays with a person.

Typical delivery
4–6 weeks
Deployment
Your contact stack
Person releases
Rights-checked
After launch
Agent Care available

Nestack Agents · Parent communication

AGT-ED-05 · Agent Care available after launch

Build a communication agent

Four questions so the call starts from your actual setup — then pick a time that suits you.

About you
Your families

No obligation. We’ll map the workflow, name the automation boundary and tell you if an agent is the wrong answer.