Sports & Fitness AI agent · Member services

Member-Services & Booking AI Agent

Answer joining, booking and billing questions from the terms already on file, book and move class places, and pass a cancellation request straight to the member-services staff who decide what is owed.

Typical delivery
4–6 weeks
Deployment
Your stack
Staff sign-off
Per-state rules
After launch
Agent Care

What this agent does

Handles the request, not what the member is owed

In

  1. 01

    Ingest the request, the member's contract terms and the club's state configuration from supported sources.

  2. 02

    Normalise membership types, add-ons and timetable references, and carry each term forward with its source.

Reason

  1. 03

    Answer booking, freeze and billing questions from the terms already on file.

  2. 04

    Apply the cancellation rules configured for the state the member contracted in.

  3. 05

    Bind each answer to a clause in the agreement, and mark what the agreement leaves open.

Decide

  1. 06

    Record a cancellation request the moment it is made, and suppress any save prompt.

  2. 07

    Route a request carrying a refund or a statutory ground to named member-services staff.

Out

  1. 08

    Retain the request, the terms it was answered from, the staff release and the confirmation.

  2. 09

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

Product statement

The agent answers and books; a member-services lead decides anything that changes what a member is owed, and the operator owns the state configuration.

Example workflow

One request, desk to release

  1. 1Request receivedApp chat, web chat, front-desk hand-off or an inbound call
  2. 2Terms assembledContract clauses, membership state, add-ons and the timetable, each with its source
  3. 3Answer proposedBooking, freeze or request record, flagged terms and confidence
  4. 4Controls appliedContract-term checks, the state's health-club and auto-renewal rules, save-prompt suppression and confidence threshold

No human action required

Stages 1 to 4 run unaided, and nothing a member is owed changes at any of them — the agent is answering, and the staff lane opens at the confidence gate.

5DecisionBranches at the confidence threshold
High confidence

Goes to member-services staff to release.

Low confidence

Adds a club-manager read first.

Staff release

What is owed is held for staff, with the terms it was read against and the confidence.

Release · Correct · Send to the club manager

Released — the member is answered

  1. 6Club systems updatedOnly where write access and approval policy allow it
  2. 7Outcome evaluatedRepeat requests, corrections made at the desk, refund disputes and complaints raised after the fact

Corrections

Each 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 what a cancellation refunds.
    • Setting the date the billing stops.
    • Judging a relocation, medical or bereavement ground.
    • Registration, bonding and the signed contract.
  • Automation boundary

    Agent acts unaided

    • ✓Book, move and release class places against the live timetable.
    • ✓Log a cancellation request the moment it is made.
    • ✓Freeze a membership inside the documented policy.
    • ✓Answer the terms already written into the member's own contract.

    Any write happens inside the boundaries agreed at implementation, never ahead of release.

    • Whether a state's cooling-off window has run.
    • Anything the member's agreement does not already say.
    • Changing a price, a term or an add-on charge.
  • Changes to state rules or release boundaries.

Example output

One request, annotated

Everything the agent answers is attached to the terms it was read from.

Member-services output · single request

Illustrative example
Request
Freeze request
Answer given
A hold on the documented freeze terms of the member's own plan
Channel
App chat, where they joined
Source of record
Membership agreement
Confidence
93%
Cancel path
Shown, and unchanged

As received

From the membership agreement and the state configuration — nothing here is decided.

Terms used

  • Membership agreement
  • Documented freeze policy
  • State configuration

Why this answer

It quotes a clause the member already signed, and asks for nothing joining did not.

What the score decides

Below the configured threshold the request picks up a club-manager read, dated when asked.

Action

ReleaseCorrectSend to the club manager

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 request

From the member's contract

03Answering

Answer from the terms

Draw on the contract on file, the timetable and the cancellation rules configured state by state — no single national rule exists.

01Approved path

Handle it at the front desk hour

Routine bookings, freezes and billing questions come back answered.

02Human review

Send staff where it costs money

Refund consequences, statutory grounds and low-confidence requests are marked, so staff time goes where being wrong is expensive.

04

Build an evidence trail

The request, the contract terms it was read against and the staff release stay together.

Integrations

Typical integrations

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

  • Club management

    • Mindbody
    • ABC Fitness
    • Glofox
    • Xplor Recreation
  • Booking and timetable

    • TeamUp
    • Legend Leisure
    • Class timetable APIs
  • Billing and payments

    • GoCardless
    • Stripe Billing
    • Direct-debit bureaux

Agent

Member services and booking

  • Reads the terms
  • Answers and books
  • Holds for staff
  • Desk and messaging

    • Zendesk
    • Intercom
    • Front-desk and app chat
  • 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 member

Each control sits inside the next. What a layer does not catch is named in the map below it.

L6 · Outermost — last line of defence

Inward → L1 · closest to the model

L6Rollback / safe modeDrop to booking-only when signals degrade, and pass requests to leave straight to the desk.Roll back
L5Version monitoringTrack model, prompt, contract-term and state-configuration changes.Track
L4TraceabilityRecord the request, the terms read, the release and the confirmation sent.Record
L3Staff releaseHolds refund-bearing requests for staff; it governs who decides, not whether the decision is right.Gate
L2Policy guardrailsTest answers against the contract terms and the state configuration; a failure returns the request.Restrict
L1Confidence thresholdsRoute low-confidence requests to a club-manager read before staff see them.Require review
Model coreAnswer produced — the booking, the freeze or the request record and confidence
L1 – L2
Test whether an answer may stand
L3
Puts what is owed in staff hands
L4 – L5
Keep the request and the terms it was answered from
L6
Drops the agent to booking-only when signals degrade

How Nestack evaluates it

Evaluate the whole request path — not only the answer the member reads.

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

Surface — the answer the member gets

  1. E1Final-output evaluationDid the answer match the clause it was drawn from?
  2. E2Step-level evaluationDid the agent use the right contract, state rules and timetable?
  3. E3Tool evaluationDid it read the correct membership and write to the correct record?
  4. E4Confidence calibrationDo low-confidence requests actually attract more staff corrections?
  5. E5Slice evaluationHow far apart do repeat rates sit across specific request types?
  6. E6Business outcomeHow many requests came back a second time, or surfaced only as a refund dispute after the member had gone?

Floor — what the operator answers for

Failure modes

Where each failure originates in the agent

Seven failure modes, each placed where it starts.

Agent lifecycle

Direction of processing →

  1. 01 · Retrieval

    1 mode

    • BG-03

      Superseded terms read

      The clause comes from an agreement the member has since replaced.

    Stage gathers

    Contract terms, state rules, timetable and billing status

  2. 02 · Reasoning

    2 modes

    • BG-04

      Wrong state's rules applied

      A member is answered against a state they did not contract in.

    • BG-06

      Save prompt on a cancel path

      A retention offer surfaces on a request to leave.

    Stage proposes

    Bookings, freezes, request records and confidence

  3. 03 · Tool / write

    2 modes

    • BG-02

      Verification added to cancelling

      A member is asked for detail joining never required.

    • BG-05

      Billing runs past the request

      A charge is taken after the request was recorded.

    Stage writes

    Only where write access and approval policy allow it

  4. 04 · Output

    1 mode

    • BG-01

      Request recorded as a hold

      A request to leave is written to the record as a freeze.

    Stage returns

    The answer the member acts on and the desk sees

  5. 05 · Change / Version

    1 mode

    • BG-07

      Silent policy regression

      A rule change narrows what the agent will pass straight through.

    Stage tracks

    Model, prompt, state rules and policy config

  • Sev-1 · a request to leave was not passed on
  • Sev-2 · a wrong answer reaches the member
  • Sev-3 · terms degrade, request routes to staff
See the complete Sports & Fitness failure-mode catalogue

Affected slices

Overall repeat rates can hide one request type

The slice the aggregate hides is cancellation under a state cooling-off window — a thin share of the queue and most of the requests a member had to raise twice. Nestack reports the repeat rate by slice, not only in total.

Slice performance — reported separately, not only in aggregate

Illustrative example
SliceFailure rateLiftLift vs. thresholdStatus
Cancellations under state cooling-off5.3%3.4×Review
Freeze and add-on requests4.0%2.6×Review
Billing disputes mid-term2.6%1.7×Watch
Class booking and rebooking0.9%0.6×Normal

Bar: repeat-request lift vs. the class-booking baseline · scale 0–4.0× · tick marks the 2.0× review threshold

2 of 4 slices over threshold

Evidence-linked improvement

The cycle closes on a test, not a note

A cycle closes when the failure is a regression case the next release has to pass. That suite is what the next request through the desk is measured against.

Improvement cycle · five stages

Switchback — the path turns at Improve and returns at Learn

  1. 01

    Detect

    Repeat requests rise in one request type.

  2. 02

    Diagnose

    The request record is where this starts: what was asked, what terms it was read against, who released it.

  3. 03

    Improve

    The change ships against a version, with the requests that exposed it attached.

  4. 04

    Verify

    Release is blocked until the affected regression cases pass again.

  5. 05

    Learn

    The case joins the permanent suite and the front-desk playbook.

  6. Learn → Detect

    The return edge. The next detection is measured 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, request workflow, evaluation, integration, then production validation and handover.

Workstream

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

  1. 01Request workflow discovery and boundary definition.Week 1
  2. 02Club-management and billing assessment.Week 1, Week 2
  3. 03Health-club, auto-renewal and ROSCA rule mapping, per state.Week 1, Week 2
  4. 04Member-record ingestion and normalisation.Week 2
  5. 05Answering logic and clause binding.Week 2, Week 3
  6. 06Confidence scoring and refund routing.Week 3
  7. 07Member-services release workflow.Week 3, Week 4
  8. 08Club-management and billing integration.Week 5
  9. 09Cancellation-path regression cases.Week 4, Week 5
  10. 10Guardrails and release controls.Week 4
  11. 11Request-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 club, one stateProductionProduction club systemsAdvancedMultiple sites / states
Introduced at Pilot
Answering from your contract terms✓✓✓
Staff release✓✓✓
Request-handling baseline✓✓✓
Introduced at Production
Reporting by membership type—✓✓
Release workflow in your systems—✓✓
Approved write-back—✓✓
Club-management-system integration—✓✓
Introduced at Advanced
Multi-state and multi-site rollout——✓
Multi-stage desk approvals——✓
High request volume——✓
Multi-site desk controls——✓
Build priceFrom $5,000From $8,000Custom quote
Final build price
Confirmed after discovery based on integrations, workflow complexity, request 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 membership agreements and their terms

    Contract-term mapping and answer configuration

    Week 1
  2. 02Representative requests from a past quarter

    Answering baseline, term extraction and clause binding

    Week 2
  3. 03The states you trade in and their cancellation rules

    Per-state cancellation-rule mapping and the automation boundary

    Week 1
  4. 04Access to relevant APIs, feeds or exports

    Club-management, booking and billing assessment, then integration setup

    Week 2
  5. 05Requests you would not want repeated

    Cancellation cases and failure-mode testing

    Week 4
  6. 06Where a request must stop and wait for staff

    Confidence scoring, refund-consequence routing, guardrails and release controls

    Week 3
  7. 07Named member-services staff to release requests

    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

Each phase sits on the weeks it actually occupies, and week 5 carries both evaluation and launch work.

Phase

W1

W2

W3

W4

W5

W6

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

Week focus

W1Request workflow discovery, term mapping and the automation boundary

W2Source integration and the answering baseline

W3Answering workflow, confidence logic and release controls

W4Evaluation suite, cancellation-path checks and failure-mode testing

W5System integration, a pilot desk queue and targeted corrections

W6A live desk queue answered under supervision, then Agent Care handover

Reading the band
A band covers the weeks its work is named in and no others. The doubled fifth week is real, not padding.
At the end of W6
Validation closes on live requests, and Agent Care picks up monitoring.
Duration
Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Sports & Fitness AI agent

Build a member-services agent around your club's own terms.

We'll map the request path, set the automation boundary and name what stops for a person. Bring your contract terms, the states you trade in and whoever signs cancellations off today.

Typical delivery
4–6 weeks
Deployment
Your club systems
Staff sign-off
Per-state rules
After launch
Agent Care available

Nestack Agents · Member services and booking

AGT-SF-01 · Agent Care available after launch

Build a member-services agent

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

About you
Your members

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