Main site

Sports & Fitness AI agent · Scouting

Scouting & Player-Evaluation AI Agent

Index the tape, score each clip against the rubric your scouts wrote, and assemble the evidence behind a name — the scout of record still decides which players go forward.

Typical delivery
4–6 weeks
Deployment
Your stack
Scout sign-off
Evidence only
After launch
Agent Care

What this agent does

Orders the evidence, not a ranking of players

In

  1. 01

    Ingest match footage, fixture data and the club's rubric from supported video, tracking or recruitment sources.

  2. 02

    Normalise fixture, minute and position labels, and carry each observation forward with the clip it came from.

Reason

  1. 03

    Index the tape into timestamped events a scout can jump straight to.

  2. 04

    Apply the evaluation rubric the club's recruitment staff wrote, criterion by criterion.

  3. 05

    Bind each rubric score to the clips that support it, and mark where the tape is thin.

Decide

  1. 06

    Flag inferences the tape cannot carry — intent, character, mentality, coachability.

  2. 07

    Route every evaluation to the scout of record for sign-off.

Out

  1. 08

    Retain the clips, the rubric scores, the scout's amendments and the sign-off on the player record.

  2. 09

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

Product statement

The agent orders evidence; the scout of record decides — and the players it passes over generate no data to correct it with.

Example workflow

One prospect, tape to sign-off

  1. 1Tape receivedMatch feed, tracking export, agency footage or a scout's upload
  2. 2Evidence assembledFixture, minute, opposition, position and prior reports, each kept with its clip
  3. 3Rubric appliedCriterion scores, supporting clips, flagged lines and confidence
  4. 4Controls appliedRubric checks, the jurisdiction rules configured for the club, prohibited-inference blocks and confidence threshold

No human action required

Stages 1 to 4 run unaided, and no player is ranked or cut at any of them — the agent orders evidence, and the scout's lane opens at the confidence gate.

5DecisionDivides at the confidence threshold
High confidence

Goes to the scout of record to sign.

Low confidence

Adds a second scout's read first.

Scout sign-off

The evaluation is held with its clips, its flagged lines and the confidence.

Accept · Amend · Send to second read

Signed — added to the report

  1. 6Recruitment record updatedOnly where write access and approval policy allow it
  2. 7Outcome evaluatedAmend rate, flagged-line outcomes, slice coverage and post-window corrections

Amendments

Every scout amendment is counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required

8 items

  • Ranking talent, or placing players in order of ability.
    • Writing an adverse note into a recruitment record.
    • Anything that reaches a contract or trade decision.
    • Cutting or removing a name from a shortlist.

  • Automation boundary

    Agent acts unaided

    • ✓Index match footage into timestamped events, searchable by fixture.
    • ✓Score each clip against the rubric the club's scouts wrote.
    • ✓Bind each criterion score to the clips that support it.
    • ✓Assemble the evidence pack, and hold it for the named scout of record.

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

    • Inferring character, mentality or coachability.
    • Extracting face geometry to identify a player.
    • Combining medical or load data into an evaluation.
  • Changes to the rubric, thresholds or approval rules.

Example output

One prospect report, annotated

Everything the agent scores is attached to the clip it was drawn from.

Evaluation output · single prospect

Illustrative example
Prospect
U18 winger, trial
Rubric line
Receives on the half-turn under pressure and releases first time
Evidence
Six clips
Source of tape
Club match footage
Confidence
89%
Reported slice
Birth quarter and maturity noted

As received

Taken from the club's footage and the clips cited — nothing on this side is inferred by the agent.

Evidence cited

  • Timestamped match clips
  • Prior scouting reports
  • Fixture and minute data

Why this score

It is what the clips show against one rubric criterion — not a view on the player.

What the score decides

Below the configured threshold the evaluation picks up a second scout's read before sign-off.

Action

AcceptAmendSend to second read

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 prospect

From the club's tape

03Evaluation

Score against the rubric

Work from the clips on file and the rubric the club's recruitment staff signed.

01Approved path

Order the tape, not the verdict

Routine rubric scoring arrives already bound to its clips.

02Human review

Send the read where tape is thin

Flagged inferences and low-confidence lines are marked, so the scout's read starts where the tape will not carry it.

04

Build an evidence trail

The evaluation, the footage it drew on and the scout who signed it stay on the report.

Integrations

Typical integrations

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

  • Video platforms

    • Hudl
    • Wyscout
    • InStat
    • Veo
    • Catapult Vision

  • Tracking data

    • Second Spectrum
    • Stats Perform
    • SkillCorner
    • Sportlogiq

  • Recruitment systems

    • Scout7
    • TransferRoom
    • SciSports
    • Twenty3

Agent

Scouting & player evaluation

  • Indexes the tape
  • Applies the rubric
  • Holds for sign-off

  • Athlete records

    • Smartabase
    • Edge10
    • Teamworks
    • Kitman Labs

  • 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 shortlist

The controls sit inside one another. What none of them catches is set out below.

L6 · Outermost — last line of defence

Inward → L1 · closest to the model

L6Rollback / safe modeNarrow the agent to tape indexing when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt, rubric and slice-configuration changes.Track
L4TraceabilityRecord the clips, the scores, the amendments and the sign-off, per player.Record
L3Scout sign-offHold evaluations for the named scout; it governs who signs, not whether the score is right.Gate
L2Policy guardrailsTest scores against the rubric and the prohibited-inference list; a failure returns the evaluation.Restrict
L1Confidence thresholdsRoute low-confidence evaluations to a second scout before sign-off.Require review
Model coreEvaluation produced — criterion scores, supporting clips, flagged lines and confidence
L1 – L2
Test whether a score may stand
L3
Puts the sign-off in a scout's hands
L4 – L5
Keep the footage the evaluation rested on
L6
Narrows to tape indexing when signals degrade

How Nestack evaluates it

Evaluate the scouting workflow — not only the finished report.

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

Surface — the report the recruitment room sees

  1. E1Final-output evaluationDid every rubric score point at a clip that supports it?
  2. E2Step-level evaluationDid the agent use the right tape, rubric version and slice configuration?
  3. E3Tool evaluationDid it read and write the correct player record and the correct field?
  4. E4Confidence calibrationDo low-confidence evaluations actually attract more scout amendments?
  5. E5Slice evaluationHow does performance change across specific prospect cohorts?
  6. E6Business outcomeHow often did the ordering simply reproduce who was already picked?

Floor — the outcome the club answers for

Failure modes

Where each failure originates in the agent

Seven failure modes, set at the stage each one starts in.

Agent lifecycle

Direction of processing →

  1. 01 · Retrieval

    1 mode

    • SQ-03

      Misattributed footage

      Clips pulled for the wrong player or the wrong fixture.

    Stage gathers

    Footage, fixture data, prior reports and the rubric

  2. 02 · Reasoning

    2 modes

    • SQ-04

      Prohibited inference

      A rubric line becomes a claim about character or mentality.

    • SQ-06

      Unsupported score

      A criterion is scored where the tape does not carry it.

    Stage proposes

    Criterion scores, supporting clips and confidence

  3. 03 · Tool / write

    2 modes

    • SQ-02

      Ordering read as ranking

      An evidence order is presented as a rank of players.

    • SQ-05

      Duplicate prospect record

      One player is evaluated twice under two records.

    Stage writes

    Only where write access and approval policy allow it

  4. 04 · Output

    1 mode

    • SQ-01

      Selection-bias echo

      Scores track who was picked before, not what the tape shows.

    Stage returns

    The evaluation the scout signs and the club acts on

  5. 05 · Change / Version

    1 mode

    • SQ-07

      Silent rubric regression

      A model or rubric change widens what the agent will assert.

    Stage tracks

    Model, prompt, rubric and slice configuration

  • Sev-1 · the agent acts outside the boundary
  • Sev-2 · a wrong score reaches the report
  • Sev-3 · tape degrades, evaluation routes to review
See the complete Sports & Fitness failure-mode catalogue

Affected slices

A healthy total hides the cohort it misses

Late maturers and Q4 births are the players academy intakes under-count, and a player never signed leaves no outcome data. Nestack reports the scout-amendment rate by slice, not only in total.

Slice performance — reported separately, not only in aggregate

Illustrative example
SliceFailure rateLiftLift vs. thresholdStatus
Q4 births and late maturers7.7%3.6×Review
Low-coverage competitions4.9%2.3×Review
Trialists with no club record3.4%1.6×Watch
Squad players on full tape1.9%0.9×Normal

Bar: scout-amendment-rate lift vs. full-tape baseline · scale 0–4.0× · tick marks the 2.0× review threshold

2 of 4 slices over threshold

Evidence-linked improvement

Where a missed evaluation ends up

The cycle ends when a case exists in the suite, not when the miss has been discussed. That suite is what the next report off the tape is measured against.

Improvement cycle · five stages

Switchback — the path turns at Improve and returns at Learn

  1. 01

    Detect

    Scout-amendment rate rises in a prospect cohort.

  2. 02

    Diagnose

    The report is the unit — read back against the clips it cites until one cause is left.

  3. 03

    Improve

    Changes carry a version and the reports that prompted them.

  4. 04

    Verify

    Release is held until the affected cases pass.

  5. 05

    Learn

    The case is kept permanently, and the evaluation rubric moves with it.

  6. Learn → Detect

    The return edge. The next miss is read 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, evaluation workflow, testing, integration, then production validation and handover.

Workstream

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

  1. 01Scouting workflow discovery and boundary definition.Week 1
  2. 02Video, tracking and recruitment assessment.Week 1, Week 2
  3. 03Evaluation rubric, slice and jurisdiction-rule mapping.Week 1, Week 2
  4. 04Footage ingestion and clip normalisation.Week 2
  5. 05Rubric logic and clip-to-score binding.Week 2, Week 3
  6. 06Confidence scoring and second-read routing.Week 3
  7. 07Sign-off and disclosure workflow.Week 3, Week 4
  8. 08Video and recruitment-system integration.Week 5
  9. 09Rank-consistency cases.Week 4, Week 5
  10. 10Guardrails and review controls.Week 4
  11. 11Evaluation-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 squad, one competitionProductionProduction recruitment systemsAdvancedMultiple leagues / regions
Introduced at Pilot
Scoring to your rubric and slices✓✓✓
Scout sign-off✓✓✓
Rank-consistency baseline✓✓✓
Introduced at Production
Reporting by cohort and birth quarter—✓✓
Sign-off workflow in your systems—✓✓
Approved write-back—✓✓
Recruitment-system integration—✓✓
Introduced at Advanced
Multi-jurisdiction rule sets——✓
Multi-stage recruitment approvals——✓
High tape volume——✓
Multi-competition scouting controls——✓
Build priceFrom $5,000From $8,000Custom 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.

  1. 01Your rubric and its criterion definitions

    Footage ingestion and clip-to-criterion mapping

    Week 1

  2. 02Representative signed evaluations

    Scoring baseline, rubric extraction and clip binding

    Week 2

  3. 03Your prohibited-inference list and target jurisdictions

    Rubric, slice and per-jurisdiction rule mapping

    Week 1

  4. 04Access to relevant APIs, feeds or exports

    Video, tracking and recruitment assessment, then integration setup

    Week 2

  5. 05Evaluations that came out inconsistent

    Consistency cases and failure-mode testing

    Week 4

  6. 06What must reach a scout before it counts

    Confidence scoring, second-read routing and per-jurisdiction controls

    Week 3

  7. 07Named scouts of record to sign evaluations

    Scout sign-off 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 follow the actual work, which is why the fifth week doubles rather than pads.

Phase

W1

W2

W3

W4

W5

W6

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

Week focus

W1Scouting workflow discovery, rubric mapping and the automation boundary

W2Source integration and the scoring baseline

W3Evaluation workflow, confidence logic and sign-off controls

W4Evaluation suite, prohibited-inference checks and failure-mode testing

W5Recruitment integration, pilot prospects and targeted corrections

W6One scouting window evaluated under the recruitment staff, then handover

Reading the band
Each bar covers the weeks its work is named in, and no others. The fifth week really carries two.
At the end of W6
The window closes validation and Agent Care owns the running agent.
Duration
Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Sports & Fitness agent

Build a scouting agent around the scouts who already sign.

Show us a scouting window, the rubric behind it and the name on the report. What would have to be true before an agent's ordering went near that signature?

Typical delivery
4–6 weeks
Deployment
Your video and scouting stack
Scout sign-off
Evidence only
After launch
Agent Care available

Nestack Agents · Scouting & player evaluation

AGT-SF-11 · Agent Care available after launch

Build a scouting agent

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

About you
Your prospects

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