Community & Class Engagement AI Agent

What this agent does

Sends the message, not the permission for it

Ingest schedules, waitlists, bookings and per-channel consent records from supported club or booking systems.

Where a member appears twice, resolve to one record and carry the consent each channel rests on.

Reason

Where a class is under-booked or a place opens, draft the nudge or the waitlist notice.

When a state caps frequency or narrows hours, the send follows that state's rule.

Where a message is promotional rather than operational, require written consent first.

Decide

When a member opts out, stop on that channel in whatever wording it arrived.

Where a junior is enrolled, message the configured guardian.

Out

Retain the message, the consent behind it, the opt-out state and the release against the member.

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

→ Product statement

The agent is configured as an operational messenger, not a companion; the studio manager decides who may be contacted and on what terms.

Example workflow

One nudge, open place to send

AgentHuman

1Class event receivedA place opens, a waitlist moves, a challenge closes or a term begins

2Consent assembledChannel consent, opt-out state, quiet hours and junior status, each with its source

3Message draftedNudge text, channel, purpose, send window and confidence

4Controls appliedConsent and purpose checks, state hour and frequency rules, junior-cohort checks and confidence threshold

No human action required

Stages 1 to 4 run unaided and nothing is sent at any of them — the agent is preparing, and the studio manager's lane opens at the confidence gate.

5DecisionDivides at the confidence threshold

High confidence

Goes to the studio manager to release.

Low confidence

Adds a programme-lead check first.

Manager release

The message is held with the consent it relies on, its purpose and the confidence.

Release · Edit · Send to the programme lead

Released — sent in the window▼

6Scheduling systems updatedOnly where write access and contact policy allow it

7Outcome evaluatedOpt-out compliance, hour and cap breaches, waitlist take-up and complaints after the send

Corrections

Manager corrections are counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required8 items

Adding or widening a member's contact permissions.

Messaging a minor outside the configured programme.

Setting how often a member may be contacted.

Starting an AI voice call — a robocall in law.

Automation boundaryAgent acts unaided

✓Draft and send class-fill and waitlist notices to consenting members.

✓Send inside the quiet-hour window configured per state.

✓Stop on the channel an opt-out arrives on, in any wording.

✓Hold promotional sends and anything to a junior for the manager.

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

Enrolling a member in a challenge they did not join.

Soliciting, filtering or routing a review by rating.

Deciding which state's rules a member falls under.

Changes to consent, frequency or junior settings.

Example output

One nudge, annotated

Everything the agent sends is attached to the permission it rested on.

Engagement output · single messageIllustrative example

Trigger

Drafted message

Purpose

Consent of record

Confidence

Send window

Waitlist place

A place has opened in Thursday's 6am class and you are next on the list

Operational

Booking consent on file

91%

Inside local quiet hours

As receivedTaken from the booking record and the consent log — nothing on this side is decided by the agent.

Permissions used Channel consent on file Opt-out state, current Waitlist position

Why this messageThe member asked for this place and a person still decides.

ActionReleaseEditSend to the programme lead

What the score decidesBelow the configured threshold the message picks up a programme-lead check.

Value

Where AI adds value

The same four claims, placed at the point in the workflow where each one applies.

Where the value landsValue 01 – 04

Every class eventFrom the booking record

03Preparation

Prepare from the record

Work from the schedule, the consent log and a scope configured as operations, not companionship.

01Approved path

Fill the class without pestering

Routine place-open and waitlist notices go out already checked.

02Human review

Keep staff on the risky sends

Promotional messages, junior programmes and low-confidence sends are marked, so the manager's time goes where the exposure is.

04Build an evidence trail

The nudge, the consent it relied on and the opt-out are kept together.

Integrations

Typical integrations

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

Club managementMindbody · ABC Glofox
Zen Planner · Club Automation

Class bookingTeamUp · WellnessLiving
Momence · Mariana Tek

MessagingTwilio · Attentive
Braze · Klaviyo

Agent

Community & class engagement

Reads the consent
Prepares the nudge
Holds for release

Member appTrainerize · Everfit
In-app and push

Observability & evaluationOpenTelemetry · 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 member

Each control encloses the last. Whatever slips past all of them is named below.

L6 · Outermost — last line of defenceInward → L1 · closest to the model

L6Rollback / safe modeCut back to scheduled class reminders when evaluation or production signals degrade.Roll back

L5Version monitoringTrack model, prompt, contact-rule and consent-configuration changes.Track

L4TraceabilityRecord the consent, the message, the send window and the release.Record

L3Manager releaseHolds promotional and junior sends for a named manager; it governs what leaves, not what happens after it has gone.Gate

L2Policy guardrailsTest each send against consent, purpose, state hours and the configured operational scope; a failure returns it.Restrict

L1Confidence thresholdsRoute low-confidence sends to a programme-lead check before the manager sees them.Require review

Model coreMessage prepared — text, channel, purpose, send window and confidence

L1 – L2Judge whether a send may go

L3Puts what leaves in a manager's hands

L4 – L5Keep the nudge and the permission behind it

L6Cuts back to scheduled reminders when signals degrade

How Nestack evaluates it

Evaluate the contact decision — not only the message that went out.

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

Surface — the message the member gets

Depth of coverage ▼

E1Final-output evaluationDid the send have consent for that channel and that purpose?

E2Step-level evaluationDid the agent use the right consent record, state rules and junior settings?

E3Tool evaluationDid it read the right member and write to the right send log?

E4Confidence calibrationDo low-confidence sends actually attract more manager corrections?

E5Slice evaluationHow far do results move across specific member segments?

E6Business outcomeHow many sends were corrected, and how many drew a complaint after delivery?

Floor — the exposure the operator answers for

Failure modes

Where each failure originates in the agent

Seven failure modes, set at the stage each one begins.

Agent lifecycleDirection of processing →

01 · Retrieval1 mode

KL-03

Stale suppression list

An opt-out recorded elsewhere has not reached the sender.

Stage gathersSchedules, waitlists, consent records and opt-out state

02 · Reasoning2 modes

KL-04

Purpose misread

A promotional offer is prepared as an operational notice.

KL-06

Companion drift

The persona slides from messenger towards a friend.

Stage proposesMessage text, channel, purpose and confidence

03 · Tool / write2 modes

KL-02

Contact outside the window

A send lands outside the hours a state allows.

KL-05

Junior contacted directly

A minor is messaged outside the configured programme.

Stage writesOnly where write access and approval policy allow it

04 · Output1 mode

KL-01

Pressure in the wording

A streak reminder reads as an obligation, not a fact.

Stage returnsWhat the member reads, and what it asks of them

05 · Change / Version1 mode

KL-07

Consent state reset

An upgrade re-enables a channel a member had switched off.

Stage tracksModel, prompt, contact rules and consent config

Sev-1 · a minor or an opt-out is contacted Sev-2 · a send breaks a state's contact rule Sev-3 · consent data degrades, send is held

See the complete Sports & Fitness failure-mode catalogue →

Affected slices

A calm total with one loud cohort inside it

Juniors and members in tighter-rule states are small cohorts that barely move the total — they are who an aggregate under-counts. Nestack reports by slice, not only in total.

Slice performance — reported separately, not only in aggregateIllustrative example

Slice	Failure rate	Lift	Status
Junior and youth programmes	4.3%	3.7×	Review
Members in tighter-rule states	3.2%	2.8×	Review
Challenge and streak messages	2.2%	1.9×	Watch
Routine class reminders	0.7%	0.6×	Normal

Bar: exception-rate lift vs. the routine-reminder baseline · scale 0–4.0× · tick marks the 2.0× review threshold 2 of 4 slices over threshold

Evidence-linked improvement

What has to exist before it closes

The cycle does not end in a post-mortem. It ends in a case the next release has to clear, and that suite is what the next send is measured against.

Improvement cycle · five stagesSwitchback — the path turns at Improve and returns at Learn

01Detect

Exception rate rises in one member cohort.

02Diagnose

Whoever answered the complaint at the desk starts here, with the send log and the consent record open.

03Improve

Each change carries a version and the messages that forced it.

04Verify

Release waits on the affected cases going green again.

05Learn

The case is kept for good, and the contact policy is rewritten around it.

Learn → DetectThe return edge. Every later run is measured 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, consent records, message workflow, evaluation, integration, then production validation and handover.

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

01Engagement workflow discovery and boundary definition.

02Booking and messaging assessment.

03Consent, state-rule and junior-cohort mapping and rule mapping.

04Schedule, waitlist and consent ingestion.

05Message assembly and consent binding.

06Confidence scoring and release routing.

07Manager release workflow.

08Booking- and messaging-platform integration.

09Contact-frequency cases.

10Guardrails and consent controls.

11Contact-trail instrumentation.

12Deployment, documentation and Agent Care handover.

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 site, one class set ProductionProduction booking systems AdvancedMultiple sites / brands

Introduced at Pilot

Sends to your consent rules✓✓✓

Manager release✓✓✓

Engagement-quality baseline✓✓✓

Introduced at Production

Reporting by member segment—✓✓

Release workflow in your systems—✓✓

Approved write-back—✓✓

Scheduling-system integration—✓✓

Introduced at Advanced

More sites and states——✓

Multi-stage release approvals——✓

High message volume——✓

Multi-site programme controls——✓

Build price From $5,000 From $8,000 Custom quote

Final build priceConfirmed after discovery based on integrations, workflow complexity, message volume, release controls and deployment requirements.

Separate from buildBuild 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 bringWe build with it

01Your consent records and where they are held → Consent and channel-permission mappingWeek 1

02Your last term of class messaging → Engagement baseline, consent binding and purpose classificationWeek 2

03Your state contact rules and junior settings → Consent, state-rule and junior-cohort mappingWeek 1

04Access to relevant APIs, feeds or exports → Booking, messaging and member-app assessment, then integration setupWeek 2

05Messages you would not want sent twice → Consent cases and failure-mode testingWeek 4

06What a nudge is never allowed to do → Confidence scoring, hold routing, guardrails and consent controlsWeek 3

07Named studio managers to release held sends → Manager release workflow, then pilot and production validationWeeks 5–6

Nothing else is required Deployment, documentation and Agent Care handover are ours.

Delivery timeline

Four phases across six weeks

The bands sit on the weeks the work occupies, which is why the fifth carries two kinds at once.

Phase W1W2W3W4W5W6

Discovery W1

Build W2 – W3

Evaluate W4 – W5

Pilot & Launch W5 – W6

Week focus W1Engagement workflow discovery, consent mapping and the automation boundary W2Booking integration and the engagement baseline W3Message workflow, confidence logic and consent controls W4Evaluation suite, contact-rule checks and failure-mode testing W5Messaging integration, a pilot term and targeted corrections W6A term of classes filled under the studio team, then Agent Care handover

Reading the bandEach band stops where its work stops. The fifth week carries two kinds because it really does.

At the end of W6The term closes the validation and Agent Care takes over monitoring.

DurationSix-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Sports & Fitness AI agent

Build an engagement agent around the consent you already hold.

Show us your class schedule, your consent records and who releases a send. Could you say today which members never consented to a promotional text, and which of them are juniors? Then we map the state rules, set the automation boundary and name what stays with a person.

Discuss this agent →

Nestack Agents · Community engagementAGT-SF-05 · Agent Care available after launch