Sports & Fitness AI agent · Anti-doping
Anti-Doping Administration AI Agent
Track the filing windows and the missing dossier elements for athletes in a registered pool, never their locations. It does not file, decide an exemption, assess a result or say a substance is permitted.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Athlete files
- Never locations
- After launch
- Agent Care
What this agent does
Tracks the window, never the whereabouts
In
- 01
A window opens each quarter, and the duty does not close with it — any change since is due immediately.
- 02
An element is missing — imaging results, a failed-alternative note — and a dossier short of one is returned unread.
Reason
- 03
A failure is dated to the first day of its quarter, so the rolling twelve months is already part spent.
- 04
A pool is read from the athlete's own organisation: UKAD's domestic pool asks less of them than its NRTP does.
- 05
An exemption is in writing before use, and a national grant may not bind the federation an athlete travels to.
Decide
- 06
An address is never held: the minimisation test a 2026 EU judgment applied, to a different question.
- 07
A notice arrives on a clock that authority sets itself; the standard calls the deadline short and never fixes it.
Out
- 08
A consent signature is held per athlete per year, with the date education reached them before their first test.
- 09
Execute write actions only inside the approval boundaries agreed during implementation.
Product statement
The agent holds the clocks and the completeness state; the athlete files personally, two named clinicians certify the dossier, and a committee decides it.
Example workflow
One filing window, open to closed
- 1Window opensPool assignment, the organisation's published calendar and the athlete's own record
- 2Position assembledWhich pool, which obligations, what the last quarter carried and what is still open
- 3State computedDue, open, filed, stale or overdue, with the deemed date behind each
- 4Controls appliedPool-rule checks, deemed-date arithmetic, element-completeness tests and a location-content block
No human action required
Stages 1 to 4 run unaided, and nothing is filed at any of them — the agent is tracking, and the programme lane opens at the completeness gate.
Goes to the athlete to act on.
Adds a programme-lead read first.
Athlete action
The position is held with its deadline, its open elements and the deemed dates.
Acknowledge · Attach element · Escalate to lead
Cleared — the athlete files it
- 6Programme record updatedOnly where write access and privacy policy allow it
- 7Outcome evaluatedPrompt delivery, element completeness, returned dossiers and what a rebuttal later found
Corrections
Every programme-lead correction is counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Filing whereabouts, or any update to one.
- Deciding whether exemption criteria are met.
- Saying whether a substance is permitted.
- Assessing a result or an analytical finding.
Automation boundary
Agent acts unaided
- ✓Track the window, the deemed dates and the update duty.
- ✓Check the dossier for each element the granting organisation lists.
- ✓Run each clock against that authority's own published interval.
- ✓Evidence that a prompt was delivered, to whom, and when.
Nothing here is filed or decided — the athlete files, clinicians sign, a panel decides.
- Storing an address, an overnight stay or a daily slot.
- Holding a credential that could file for an athlete.
- Deciding which pool an athlete belongs in.
- Ranking athletes by predicted non-compliance.
Example output
One filing window, annotated
In its 2025 recommendations the EDPB called the proposed retention very long for data that may include geolocations and minors, so this side holds a date and a state, never an address.
Whereabouts position · single window
Illustrative example- Athlete
- Registered-pool athlete
- Obligation
- Quarterly filing, plus every update the quarter has since made necessary
- Deemed date
- First day of quarter
- Source of record
- Pool assignment, read-only
- Confidence
- Update due
- Who files
- The athlete, personally
As received
Taken from the pool assignment and the organisation's calendar — no location content is held on this side.
What the position shows
- The window and its dates
- The elements still open
- The prompts delivered
Why this stays open
Filing on the deadline opens the duty rather than discharging it.
What the score decides
Below the configured threshold the position picks up a programme-lead read before.
Action
AcknowledgeAttach elementEscalate to leadValue
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 window
From the pool assignment
Tracks duties, not substances
Our nutrition agent suppresses a supplement question and routes it to a named dietitian; this one records that it was asked and to whom.
It holds clocks, not places
Minimisation asks whether a less restrictive means exists. A due date, a state and an element flag are that means.
Where the exposure actually sits
Not on the programme — in CAS 2020/A/7526 & 7559 an athlete who delegates filing assumes all risks for the third party's errors. So this agent never files.
Build an evidence trail
The window, the element still missing and the clinician who signed stay on the dossier.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
Anti-doping systems
- ADAMS
- read-only
- Organisation portals
Programme calendar
- Competition and camps
- Travel approvals
- fixtures
Athlete management
- Smartabase
- Teamworks
- Kitman Labs
- Edge10
Agent
Anti-doping administration
- Reads the pool
- Tracks the window
- Holds for the athlete
Education and consent
- LMS
- module records
- Consent-form records
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 athlete
Six layers, each drawn tighter than the last. What the stack misses is listed in the map below.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether a position may stand
- L3
- Puts the filing in the athlete's hands
- L4 – L5
- Keep the window and the element behind it
- L6
- Narrows to dossier assembly when signals degrade
How Nestack evaluates it
Evaluate the whole obligation position — not only the date on the calendar.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the position the athlete acts on
- E1Final-output evaluationDid the position carry the deemed dates the rolling window is computed from?
- E2Step-level evaluationDid the agent use the right pool, the right authority and the right interval?
- E3Tool evaluationDid it read the correct athlete and the correct quarter?
- E4Completeness calibrationDo dossiers passed as complete actually avoid being returned unread?
- E5Slice evaluationHow does performance change across specific pool and obligation types?
- E6Business outcomeHow many positions needed a correction before the athlete acted?
Floor — the exposure the athlete carries
Failure modes
Where each failure originates in the agent
Seven failure modes, each placed at the stage where the fault starts.
Agent lifecycle
Direction of processing →
01 · Retrieval
1 mode
- IA-03
Stale pool assignment
An athlete moved pool and the old rule set held.
Stage gathers
Pool assignment, calendars and published intervals
02 · Reasoning
2 modes
- IA-04
Notice-date arithmetic
The rolling window is run from the notice, not the deemed date.
- IA-06
Cross-regime mismatch
A collegiate exception is tracked as though it were a TUE.
Stage proposes
The window, its deemed dates and open elements
03 · Tool / write
2 modes
- IA-02
Location content retained
A rebuttal pack outlives the response it was built for.
- IA-05
Green on a stale filing
A quarter filed on time is marked done and left there.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- IA-01
Element passed as present
A dossier goes to signature without its clinical evidence.
Stage returns
The position a named athlete acts on personally
05 · Change / Version
1 mode
- IA-07
Silent rule-set drift
An authority changes an interval and the register does not.
Stage tracks
Model, prompt, pool rules and verification dates
- Sev-1 · the agent acts outside the boundary
- Sev-2 · a wrong date reaches the athlete
- Sev-3 · source degrades, window holds open
Affected slices
The squad total conceals one pool
A squad-level element-completeness figure can sit inside tolerance while a single pool carries nearly all of the rework. Nestack reports the correction rate by pool, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Registered testing pool | 7.1% | 3.6× | Review | |
| Exemption dossiers | 5.2% | 2.7× | Review | |
| Results-management clocks | 3.7% | 1.9× | Watch | |
| Settled quarterly filings | 2.5% | 1.3× | Normal |
Bar: correction-rate lift vs. settled-quarterly-filing baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
The window was running already
A cycle is done when the backdated failure has become a case the next release must pass. That suite is what the next window worked is measured against.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Correction rate rises in one pool.
- 02
Diagnose
The quarter that was already half spent when the notice arrived is read back until one cause remains.
- 03
Improve
Stamp the change; the windows behind it are filed against that number.
- 04
Verify
Each touched window case is run again, and one red stops the release.
- 05
Learn
It stays a standing test, and the pool rules move with it.
Learn → Detect
The return edge. The next window is worked 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, obligation tracking, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Obligation-workflow discovery and boundary setting.Week 1
- 02Pool and authority register assessment.Week 1, Week 2
- 03Window, element and clock rule mapping per authority.Week 1, Week 2
- 04Calendar ingestion and normalisation.Week 2
- 05Deemed-date logic and element binding.Week 2, Week 3
- 06Completeness scoring and exception routing.Week 3
- 07Athlete prompt and escalation workflow.Week 3, Week 4
- 08Programme-calendar and roster integration.Week 5
- 09Window and element cases.Week 4, Week 5
- 10Guardrails and privacy controls.Week 4
- 11Window-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 squad, one pool | ProductionProduction programme systems | AdvancedMultiple sports / organisations |
|---|---|---|---|
| Introduced at Pilot | |||
| Tracking to your pools and clocks | ✓ | ✓ | ✓ |
| Athlete-action gating | ✓ | ✓ | ✓ |
| Element-completeness baseline | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Reporting by pool | — | ✓ | ✓ |
| Review workflow in your systems | — | ✓ | ✓ |
| Approved write-back | — | ✓ | ✓ |
| Programme-calendar integration | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Complex multi-authority rule sets | — | — | ✓ |
| Multi-stage programme escalation | — | — | ✓ |
| High athlete volume | — | — | ✓ |
| Multi-organisation window controls | — | — | ✓ |
| Build price | From $5,000 | From $8,000 | Custom quote |
- Final build price
- Confirmed after discovery based on integrations, workflow complexity, transaction 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.
You bring
We build with it
01Your pool assignments and athlete roster
→Pool and authority register setup
Week 102Representative dossiers and past windows
→Completeness baseline and deemed-date binding
Week 203Your authority intervals and their verification dates
→Window, element and clock rule mapping, and the boundary
Week 104Access to relevant APIs, feeds or exports
→Calendar, roster and LMS assessment, then integration setup
Week 205Dossiers you would not want refused
→Rebuttal cases and the evaluation suite
Week 406What no dossier may leave open
→Completeness scoring, exception routing, guardrails and privacy controls
Week 307Named programme leads, and the athletes prompted
→Athlete prompt 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
What sizes each band is elapsed work and not the page, and so the fifth week accordingly doubles.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Obligation discovery, rule mapping and the automation boundary
W2Source integration and the completeness baseline
W3Window tracking, deemed-date logic and privacy controls
W4Evaluation suite, rebuttal cases and failure-mode testing
W5Programme-system integration, pilot pools and targeted corrections
W6One filing quarter run under the programme lead, then Agent Care handover
- Reading the band
- Each bar covers the weeks its own work is named in and no others. The fifth week genuinely carries two.
- At the end of W6
- When the quarter 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 · Sports & Fitness agent
Build an anti-doping agent around the athlete who has to file.
Tell us which pools your athletes sit in and which organisation governs each. Not a filing service. A clock, an element list and a prompt you can evidence — for elite programmes, governing bodies and collegiate compliance offices, not clubs.
- Typical delivery
- 4–6 weeks
- Deployment
- Your pools and calendars
- Athlete files
- Never locations
- After launch
- Agent Care available
Nestack Agents · Anti-doping administration
AGT-SF-19 · Agent Care available after launch