Media & Entertainment AI agent · Royalty admin
Music Royalty Administration AI Agent
Maintain the identifiers and ownership data royalty payments depend on, reconcile inbound statements against registrations, and assemble the evidence behind a claim or a report of usage a named person signs.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- A person signs
- Before signing
- After launch
- Agent Care
What this agent does
Maintains the data, never the certification
In
- 01
A work is registered, and its writer, publisher and work identifiers are read from the registration on file.
- 02
A recording is matched to a work by identifier, never by a title string two compositions happen to share.
Reason
- 03
A work whose shares do not sum, or whose writer resolves to more than one party code, is held back.
- 04
A statement arrives, and each line is reconciled against expected usage with the variance written down.
- 05
A statement line is tested at the rate in force to the end of 2027; the next period's settlement is not adopted.
Decide
- 06
A report of usage is assembled to the identifier content 37 CFR 210.27 prescribes, and gaps are named.
- 07
A payload leaves against a supported message standard, since older versions of those standards were sunset.
Out
- 08
A work is retained with what it was matched from — the registration, the statement and the confidence.
- 09
Execute write actions only inside the approval boundaries agreed during implementation.
Product statement
The agent maintains data and reconciles; a duly authorised officer of the licensee certifies the report of usage, and a named administrator claims the work.
Example workflow
One work, registration to signature
- 1Registration receivedCatalogue registration, splits, identifiers and prior statements
- 2Identifiers resolvedRecording and work codes, writer and publisher codes, each with its source
- 3Match proposedWork, shares, the matching basis and confidence
- 4Controls appliedShare-sum checks, identifier-resolution checks, earnings-tier checks and confidence threshold
No human action required
Stages 1 to 4 run unaided, and nothing is claimed or certified at any of them — the agent is matching, and the human lane opens at the confidence gate.
Goes to the royalty manager to assert.
Adds a rights-research read first.
Royalty manager review
The match is held with its basis, its unresolved identifiers and the confidence.
Assert · Amend · Send to rights research
Asserted — released to claim
- 6Registration systems updatedOnly where write access and approval policy allow it
- 7Outcome evaluatedAmendments, ownership conflicts raised, unmatched carry-over and distributions received
Amendments
Every manager amendment is counted, including one made after a distribution has already run.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Certifying a report of usage or its annual examination.
- Asserting ownership of a share of a work.
- Deciding an ownership conflict between two claimants.
- Agreeing or amending writer and publisher splits.
Automation boundary
Agent acts unaided
- ✓Maintain the identifier and ownership data behind the catalogue.
- ✓Propose a recording-to-work match with its basis and confidence.
- ✓Reconcile inbound statements and write the variance down.
- ✓Assemble the report content the rule prescribes,.
Any write happens inside the boundaries agreed at implementation, never ahead of a signature.
- Filing or transmitting a report to the collective.
- Deciding to escalate a discrepancy to an audit.
- Accepting a distribution as correct and final.
- Changes to matching, reporting or approval rules.
Example output
One work in the catalogue, annotated
Everything the agent matches is attached to the registration it was drawn from.
Matching output · single work
Illustrative example- Catalogue
- Back-catalogue single
- Proposed match
- Recording resolved to the registered work by code, not by title
- Earnings tier
- Lowest earning tier
- Source of record
- Registration on file
- Confidence
- 88%
- Shares
- Writer and publisher, from the registration
As received
Taken from the catalogue registration and the statement on file — nothing on this side is written by the agent.
Evidence used
- Recording identifier
- Registered work code
- Prior statement line
Why this match
The recording resolves to one registered work and the shares behind it sum.
What the score decides
Below the configured threshold the match picks up a rights-research read before it.
Action
AssertAmendSend to rights researchValue
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 work
From the registration
Match on identifiers
Resolve each recording to the work by code, never by title. The collective holding that database had its designation continued in June 2026, with no conditions imposed.
Small works match worst
Matching is tiered by what a work earns. The top tier matches almost perfectly and the bottom tier does not, and the bottom tier is where nearly all the works are.
The old black box is not the pitch
The historical unmatched pool still quoted in the market was revised down and has largely been matched and paid. The first market-share distribution of what remains is planned, not begun.
Build an evidence trail
The work, the registration it was matched to and the officer who certified 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.
Collective and societies
- Mechanical licensing collective
- PRO and society statements
Registration and identifiers
- Work and recording codes
- Party and name codes
Publishing administration
- Royalty statement systems
- Catalogue and splits
Agent
Music royalty administration
- Reads the registration
- Proposes the match
- Holds for signature
Reporting and delivery
- DDEX message standards
- Distributor statement feeds
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 report
Six checks folded one into another. What reaches the far side is named in the map below this.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Test whether a match may stand
- L3
- Leaves the signature with a named person
- L4 – L5
- Keep the work and the registration behind it
- L6
- Falls back to identifier reporting when signals degrade
How Nestack evaluates it
Evaluate the matching workflow — not only the finished match.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the claim the collective receives
- E1Final-output evaluationDid the recording resolve to the work the registration names?
- E2Step-level evaluationDid the agent use the current registration, splits and identifiers?
- E3Tool evaluationDid it read the correct catalogue and write the correct field?
- E4Confidence calibrationDo low-confidence matches actually attract more manager amendments?
- E5Slice evaluationHow does performance change across specific earnings tiers?
- E6Business outcomeHow many works stayed unmatched, or were corrected after a distribution?
Floor — the payment the writer receives
Failure modes
Where each failure originates in the agent
Seven failure modes, placed at the stage each one originates.
Agent lifecycle
Direction of processing →
01 · Retrieval
1 mode
- JQ-03
Stale registration read
Splits are read from a superseded registration.
Stage gathers
Registrations, identifiers, splits and statements
02 · Reasoning
2 modes
- JQ-04
Title-string match
A recording matches on a title two works share.
- JQ-06
Duplicate party codes
One writer resolves to more than one code.
Stage proposes
Work, shares, the matching basis and confidence
03 · Tool / write
2 modes
- JQ-02
Match taken as a claim
A proposed match is treated as an assertion.
- JQ-05
Work registered twice
The same work is registered in two places.
Stage writes
Only where write access and approval policy allow it
04 · Output
1 mode
- JQ-01
Work left unmatched
Royalties accrue against no identified owner.
Stage returns
The match a distribution is later paid against
05 · Change / Version
1 mode
- JQ-07
Silent matching drift
A model or rule change alters what is matched.
Stage tracks
Model, prompt, matching rules and report formats
- Sev-1 · a claim asserted past the boundary
- Sev-2 · a wrong match reaches a distribution
- Sev-3 · source degrades, match goes to review
Affected slices
An aggregate match rate flatters a catalogue
A catalogue-level match rate is set by the works at the top of it, where matching is close to complete. The tiers beneath carry the failure, and Nestack reports the unmatched rate by earnings tier, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Lowest-earning works | 8.2% | 3.7× | Review | |
| Cover versions and re-titles | 5.9% | 2.7× | Review | |
| Reissues and compilations | 3.8% | 1.7× | Watch | |
| Highest-earning works | 1.1% | 0.5× | Normal |
Bar: unmatched-rate lift vs. highest-earning baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
Nothing closes until a test exists
A cycle shuts when the unmatched work is a case the next release has to close. That suite is what the next report certified is measured against.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Unmatched rate rises in an earnings tier.
- 02
Diagnose
The work that earned everywhere and matched almost nowhere is traced back to one cause — a re-title that broke its link to the composition.
- 03
Improve
The correction ships numbered, with the works that prompted it attached.
- 04
Verify
While a touched match case is failing, the release does not move.
- 05
Learn
It is kept as a standing case, and the identifier rules are revised alongside.
Learn → Detect
The return edge. The next detection runs against a suite one unmatched-work 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, matching workflow, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Royalty workflow discovery and boundary definition.Week 1
- 02Registration and statement source review.Week 1, Week 2
- 03Identifier, split and report-content rule mapping.Week 1, Week 2
- 04Registration ingestion and normalisation.Week 2
- 05Matching logic and identifier binding.Week 2, Week 3
- 06Confidence scoring and flag routing.Week 3
- 07Certification and claim workflow.Week 3, Week 4
- 08Royalty and reporting-system integration.Week 5
- 09Match and identifier cases.Week 4, Week 5
- 10Guardrails and reporting controls.Week 4
- 11Work-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 catalogue, one source | ProductionProduction royalty systems | AdvancedMultiple societies / territories |
|---|---|---|---|
| Introduced at Pilot | |||
| Matching to your registrations | ✓ | ✓ | ✓ |
| Human certification | ✓ | ✓ | ✓ |
| Match-rate baseline | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Reporting by catalogue | — | ✓ | ✓ |
| Approval workflow in your systems | — | ✓ | ✓ |
| Approved write-back | — | ✓ | ✓ |
| Registration-system integration | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Multi-society rule sets | — | — | ✓ |
| Multi-stage claim approvals | — | — | ✓ |
| High catalogue volume | — | — | ✓ |
| Multi-society matching 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 catalogue registrations and split records
→Registration ingestion and identifier mapping
Week 102Statements you have already received
→Matching baseline and share-binding checks
Week 203Your reporting obligations and their deadlines
→Identifier, split and report-content rule mapping
Week 104Access to relevant APIs, feeds or exports
→Registration and statement assessment, then integration setup
Week 205Reports you would not want examined
→Unmatched cases and the evaluation suite
Week 406What no report may certify
→Confidence scoring, flag routing, guardrails and reporting controls
Week 307Named royalty managers and the certifying officer
→Certification 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 band lands on the weeks the work really occupies, so the fifth carries two of them together.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Royalty workflow discovery, rule mapping and the automation boundary
W2Registration and statement integration and the matching baseline
W3Matching workflow, confidence logic and reporting controls
W4Evaluation suite, share and identifier checks and failure-mode testing
W5Reporting-format integration, a pilot catalogue and targeted corrections
W6One accounting period run under the royalty manager, then Agent Care handover
- Reading the band
- A bar spans the weeks its work is named in and no others. The week 5 overlap is work, not padding.
- At the end of W6
- Validation closes on live statements, and monitoring moves to Agent Care.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Media & Entertainment AI agent
Build a royalty administration agent around the data a claim rests on.
Show us your registrations, your statements and who signs. The certification that carries a personal signature sits with the blanket licensee's officer on each report of usage, and with the CPA who examines the annual one.
- Typical delivery
- 4–6 weeks
- Deployment
- Your registration stack
- A person signs
- Before signing
- After launch
- Agent Care available
Nestack Agents · Music royalty administration
AGT-ME-22 · Agent Care available after launch