Travel & Hospitality AI agent · ATOL protection

ATOL Financial Protection AI Agent

Sift every booking for whether it is licensable under SI 2012/1017, tie each protection contribution to the booking that raised it, and hold the return for the ATOL Reporting Accountant who signs it.

Typical delivery
4–6 weeks
Deployment
Your stack
Named ARA
One signature
After launch
Agent Care

What this agent does

Assembles the schedules, never the signature

In

  1. 01

    A booking is taken, and reg 9 of SI 2012/1017 decides whether flight accommodation was made available.

  2. 02

    A booking is licensable, and the per-passenger protection contribution attaches to it, not to the traveller.

Reason

  1. 03

    A reporting period closes, and the contribution reaches the Air Travel Trust inside six weeks.

  2. 04

    A remittance is made, and the Contributions to the Air Travel Trust Regulations 2007 govern it.

  3. 05

    A renewal date falls, and the application and the accountants report are due 1 August or 1 February.

Decide

  1. 06

    A licence expires ungranted, and the CAA publishes your name and licensable business must stop.

  2. 07

    A package is sold, and SI 2018/634 runs beside the ATOL licence rather than in place of it.

Out

  1. 08

    A schedule is requested, and only an individually registered reporting accountant may sign for it.

  2. 09

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

Product statement

The agent classifies bookings and tracks the clocks. A registered ATOL Reporting Accountant signs the Annual Accountants Report; the licence holder remits.

Example workflow

One booking, evidence to signature

  1. 1Booking evidence receivedReservation exports, trust-account entries, contribution returns or trading projections
  2. 2Booking context assembledThe booking, whether it is licensable, the licence it sits under and the period it falls in
  3. 3Return evidence draftedThe bookings classified, the contributions tied to them, the gaps and completeness
  4. 4Controls appliedRenewal-clock checks, remittance-clock checks, classification checks and completeness confidence

No human action required

Stages 1 to 4 run unaided, and nothing is signed at any of them — the agent is assembling, and the finance lane opens at the completeness gate.

5DecisionSplits at the completeness gate
Evidence sufficient

Goes to the ATOL Reporting Accountant to sign.

Anything thin

Adds a finance read first.

Finance review

The return is held with its bookings, its contributions and the schedules behind them.

Sign · Append evidence · Send to finance review

Signed — by a registered ARA

  1. 6Booking and contribution records updatedOnly where write access and records policy allow it
  2. 7Outcome evaluatedClassification accuracy, contribution coverage, reviewer corrections and what the read found

Corrections

Each finance correction is counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required

8 items

  • Signing the Annual Accountants Report.
    • Certifying a renewal application to the CAA.
    • Remitting the contribution to the Trust.
    • Deciding that a booking is not licensable.
  • Automation boundary

    Agent acts unaided

    • ✓Assemble the schedules a reporting accountant is going to request.
    • ✓Track the renewal clock and the remittance clock.
    • ✓Tie every protection contribution to its own booking.
    • ✓Flag the booking whose licensable status the records cannot settle.

    Nothing is signed or remitted except by a named person, inside the agreed boundaries.

    • Judging whether a return may be signed.
    • Telling the CAA that a licence is in order.
    • Setting the licence type a holder trades under.
  • Changes to bookings, returns or trust records.

Example output

One booking, annotated

The renewal peaks fall on 1 August and 1 February; this record is what a single booking carried into the return.

Return evidence · single booking

Illustrative example
Booking
Flight-inclusive package, reg 9
Recorded as
Classified licensable, contribution attached
Licence class
Standard ATOL
Evidence of record
Booking record, 3 August 2026
Confidence
Held unsigned
Held for
A registered ATOL Reporting Accountant

As received

Taken from the reservation export and the trust-account entry — it reaches as far as those sources do.

What the record holds

  • Booking record
  • Contribution return
  • Trust-account entry

Why no signature here

Whether the return may be signed is an act reserved to a registered ARA.

What the score decides

Below the configured threshold the return picks up a finance read before the ARA sees it.

Action

SignAppend evidenceSend to finance review

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 booking

From the system that sold it

03Evidence

Where the evidence is used

Our group sales agent signs a contractual response to a planner from an approved rate library; this is a licence condition enforced by a regulator that can stop you trading.

01Approved path

The accountant must be named

A firm cannot stand in for the person: "All ATOL reporting needs to be signed by an ARA," and the ARA registers individually with ACCA, AIA, ICAEW, ICAI, ICAS or IFA.

02Human review

What was checked, and not found

The timing wording of reg 17, the accountants report deadline expressed in months after the accounting reference date, and which ATOL Standard Term governs that report were all checked and none was confirmed.

04

Build an evidence trail

The booking, the contribution it attracted and the accountant who signed stay on the return.

Integrations

Typical integrations

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

  • Reservation and booking

    • Booking engine
    • PMS
    • Tour operator systems
  • Finance and ledger

    • Management accounts
    • Revenue and passenger splits
  • Trust and bonding

    • Trust-account statements
    • Bond and insurance evidence

Agent

ATOL financial protection

  • Reads the bookings
  • Assembles the return
  • Holds for the accountant
  • Contributions and returns

    • Portal exports
    • returns
    • Air Travel Trust remittances
  • Observability & evaluation

    • OpenTelemetry
    • Langfuse
    • Supported monitoring/evaluation sources

Integration availability depends on the client's existing systems and API access.

Agent controls

Six sifts between the model and the accountant

Six sifts over the same file, each one finer than the last. Whatever remains is set out in the map below.

L6 · Outermost — last line of defence

Inward → L1 · closest to the model

L6Rollback / safe modeNarrow the agent to evidence assembly when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt and licence rules; as of 24 August 2026 the scheme stands unchanged, and ATIPAC reports that the preferred-options consultation on reform has not yet happened.Track
L4TraceabilityRecord each booking, the contribution it attracted, the schedule it feeds and every read of the return.Record
L3Accountant releaseHold the return for a registered ATOL Reporting Accountant; the hold governs release, not whether the figures beneath it are right.Gate
L2Scope guardrailsTest the evidence against SI 2012/1017 and the Contributions to the Air Travel Trust Regulations 2007 as configured.Restrict
L1Confidence thresholdsRoute a thin return to a finance read first; reg 19 leaves the form and content of an ATOL Certificate to CAA publication, so the binding spec moves.Require review
Model coreEvidence assembled — the bookings, the contributions, the schedules and completeness
L1 – L2
Test whether a return may stand
L3
Puts the signature in a person's hands
L4 – L5
Keep the booking and the contribution behind it
L6
Holds the return unsigned when signals degrade

How Nestack evaluates it

Evaluate the whole assembly — not only the return evidence that comes out.

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

Surface — the return the CAA reads

  1. E1Final-output evaluationDid the evidence record what each booking actually attracted?
  2. E2Step-level evaluationDid the agent read the right licence, the right period and the live booking record?
  3. E3Tool evaluationDid it read and write the correct booking and the correct contribution?
  4. E4Confidence calibrationDo low-confidence returns actually attract more finance corrections?
  5. E5Slice evaluationHow does performance change across specific booking classes?
  6. E6Business outcomeHow many returns needed a correction before the accountant signed?

Floor — the licence the holder answers for

Failure modes

Where each failure originates in the agent

Seven failure modes, each set at the stage where it first appears.

Agent lifecycle

Direction of processing →

  1. 01 · Retrieval

    1 mode

    • JW-03

      Stale booking read

      The reservation read is not the one now on the licence.

    Stage gathers

    The bookings, the licence, the periods and the dates

  2. 02 · Reasoning

    2 modes

    • JW-04

      Booking asserted, not shown

      A booking is called licensable without its record.

    • JW-06

      Stalled reform read as live

      Reform proposals are worked as though in force.

    Stage proposes

    Bookings, their contributions and completeness

  3. 03 · Tool / write

    2 modes

    • JW-02

      Thin return passed forward

      A return moves on without the finance read.

    • JW-05

      Bound to the wrong booking

      A remittance is filed against the wrong booking.

    Stage writes

    Only where write access and approval policy allow it

  4. 04 · Output

    1 mode

    • JW-01

      Signed, evidence unrecorded

      The return shows a signature but not what supported it.

    Stage returns

    The return an accountant signs and the CAA reads

  5. 05 · Change / Version

    1 mode

    • JW-07

      Silent clock regression

      A configuration change moves the deadline, not the return.

    Stage tracks

    Model, prompt, licence rules and return fields

  • Sev-1 · a return signed on no evidence
  • Sev-2 · wrong evidence reaches the return
  • Sev-3 · source degrades, return holds unsigned
See the complete Travel & Hospitality failure-mode catalogue

Affected slices

Flight-inclusive packages absorb the corrections

A licence-level contribution-completeness figure can read clean while flight-inclusive packages carry most of the rework. Nestack reports the correction rate by booking class, not only in total.

Slice performance — reported separately, not only in aggregate

Illustrative example
SliceFailure rateLiftLift vs. thresholdStatus
Flight-inclusive packages10.9%3.7×Review
Agency sales for another holder7.8%2.6×Review
Dynamic and split-contract sales4.8%1.6×Watch
Seat-only sales outside the licence2.3%0.8×Normal

Bar: correction-rate lift vs. seat-only baseline · scale 0–4.0× · tick marks the 2.0× review threshold

2 of 4 slices over threshold

Evidence-linked improvement

What a missed renewal costs

A cycle shuts when the unremitted contribution is a regression case. That suite is what the next return assembled is measured against.

Improvement cycle · five stages

Switchback — the path turns at Improve and returns at Learn

  1. 01

    Detect

    Correction rate rises on flight-inclusive packages.

  2. 02

    Diagnose

    The renewal date that arrived while the accountant was still waiting on a schedule is read back until one cause remains.

  3. 03

    Improve

    The change ships numbered, and the bookings that forced it ride with it.

  4. 04

    Verify

    Nothing releases while one touched booking case is still red.

  5. 05

    Learn

    It is retained for good, and the return rules are amended in that same commit.

  6. Learn → Detect

    The return edge. The next licence year 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, return assembly, evaluation, integration, then production validation and handover.

Workstream

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

  1. 01Licensable-booking discovery and boundary work.Week 1
  2. 02Booking, ledger and trust sources.Week 1, Week 2
  3. 03Booking-to-contribution and renewal-clock mapping.Week 1, Week 2
  4. 04Booking and contribution ingestion.Week 2
  5. 05Booking, licence and record binding.Week 2, Week 3
  6. 06Completeness scoring and review routing.Week 3
  7. 07Reporting accountant signing workflow.Week 3, Week 4
  8. 08Booking and ledger-system integration.Week 5
  9. 09Contribution and certificate cases.Week 4, Week 5
  10. 10Guardrails and signing controls.Week 4
  11. 11Return-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 licence, one yearProductionProduction renewal workflowAdvancedMultiple licences / entities
Introduced at Pilot
Return assembly to your bookings✓✓✓
Reporting accountant release✓✓✓
Licensable-booking baseline✓✓✓
Introduced at Production
Reporting by licence—✓✓
Signing workflow in your systems—✓✓
Approved write-back—✓✓
Booking-system integration—✓✓
Introduced at Advanced
Multi-licence groups——✓
Cross-period evidence packs——✓
Large booking registers——✓
Multi-deadline renewal controls——✓
Build priceFrom $5,000From $8,000Custom quote
Final build price
Confirmed after discovery based on integrations, workflow complexity, booking volume, approval 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 licences and the bookings sold under each

    Booking inventory mapping and evidence capture

    Week 1
  2. 02Representative booking, ledger and trust records

    Record binding, contribution logic and the return baseline

    Week 2
  3. 03Your classification rules under reg 9 and reg 10

    Booking mapping, contribution binding and the automation boundary

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

    Booking, ledger and trust-source assessment, then integration setup

    Week 2
  5. 05Returns you would not want reviewed

    Contribution cases and failure-mode testing

    Week 4
  6. 06What no accountants report may establish

    Completeness scoring, review routing, guardrails and release controls

    Week 3
  7. 07A registered ATOL Reporting Accountant to sign

    Signing 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

The bands below are weeks of genuine effort rather than spacing, so one of them must carry a pair.

Phase

W1

W2

W3

W4

W5

W6

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

Week focus

W1Licence workflow discovery, booking mapping and the automation boundary

W2Source integration and the contribution-completeness baseline

W3Return assembly, contribution logic and release controls

W4Evaluation suite, classification cases and failure-mode testing

W5Ledger integration, pilot returns and targeted corrections

W6One licence year run under the accountable director, then Agent Care handover

Reading the band
Each bar covers only the weeks its own work is named for. The fifth holds two because the work does.
At the end of W6
Validation closes on live returns, and Agent Care picks up the watch.
Duration
Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Travel & Hospitality AI agent

Build an ATOL agent around the return a registered accountant has to sign.

Show us one booking and the contribution it attracted. If your licence expires on 30 September, then the renewal application and the accountants report fall due by 1 August, and an ungranted licence puts your name on a published list. Package travel duties are a different lane.

Typical delivery
4–6 weeks
Deployment
Your bookings
Named ARA
One signature
After launch
Agent Care available

Nestack Agents · ATOL protection

AGT-TH-15 · Agent Care available after launch

Build an ATOL agent

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

About you
Your bookings

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