Media AI agent · Promotion
Marketing-Content Generation AI Agent (Title Promotion)
Assemble trailers, cutdowns, key art and synopses from the cut and stills a person cleared — with spoilers and embargoes held, no generated likeness or voice, and a named marketing lead releasing the asset.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Release
- Marketing lead
- After launch
- Agent Care
What this agent does
Builds the promo from cleared material
In
- 01
Take the promo brief — the title, the beat it sells, the surface, the territory and the date it may go out.
- 02
Read the approved cut, the EPK stills, the cue sheet and the consents and approvals held against each element.
Reason
- 03
Assemble trailer, teaser and cutdown timelines from the cleared selects, and keep the frames each shot came from.
- 04
Draft synopses, episode blurbs and social copy at the length each surface takes, and mark what they give away.
- 05
Lay out key art from approved stills, inside the billing block and the artwork approvals the contracts require.
Decide
- 06
Hold anything that shows a spoiler, runs ahead of an embargo, or exceeds the rating the promo carries.
- 07
Route every music cue, archive clip and use of a performer's likeness to the person who clears that one.
Out
- 08
Hand the marketing lead the asset with the material it used and the approval behind each element.
- 09
Retain the cut version, the stills and cues used, the rights read, the lead's recut and where the asset ran.
Product statement
The agent assembles from material a person cleared. It creates no digital replica of a performer's face or voice, and nothing it builds goes out until a named marketing lead releases it.
Example workflow
One promo asset, end to end
- 1Promo brief receivedThe title, the beat it sells, the surface, the territory and the embargo it sits behind
- 2Cleared material gatheredThe approved cut and select reels, EPK stills, the cue sheet, talent consents and the artwork approvals on file
- 3Asset assembledCutdown, key art or synopsis built from cleared selects, each element carrying the frame, still or cue it came from
- 4Rules appliedSpoiler and embargo rules, the rating the promo carries, the territory it runs in, and billing and credit obligations
No human action required
Stages 1 to 4 run without a person in the loop — the material is gathered and the asset assembled before anyone is asked to look at it. An element with no consent or licence behind it ends that stretch.
Goes to the marketing lead to release.
Stops with the person who clears it.
Marketing lead
Reads the asset beside the material it was cut from, the approval behind each element and what it gives away.
Release · Recut · Send for clearance
Released — handed back
- 6Released by the marketing leadScheduled to the channel only after a named marketing lead releases it; the agent posts nothing itself
- 7Outcome reviewedAssets recut before release or pulled after it, spoiler and embargo escapes, rights holds, by title and surface
Recuts
Edits the lead makes before release are counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Authorising a generated or altered performer likeness.
- Clearing music, archive clips or talent usage.
- Releasing an asset to a channel or partner.
- Deciding what may be shown before an embargo lifts.
Automation boundary
Agent acts unaided
- ✓Assemble cutdowns, key art and synopses from cleared selects.
- ✓Draft copy at the length each surface actually takes.
- ✓Mark the spoilers, embargo dates and rights gaps it finds.
- ✓Carry the consent, licence and approval behind each element.
Write actions run only inside the approval boundaries agreed during implementation. The release is not one of them.
- Approving key art against billing and credit terms.
- Setting the rating or territory a promo may carry.
- Confirming an unannounced date, casting or plot point.
- Changing spoiler rules, embargo dates or approval routes.
Example output
One promo asset, annotated
Everything the agent assembles is attached to the cut and the approvals it came from.
Promo output · single asset
Illustrative example- Asset requested
- Episode 4 promo, 30 seconds
- Cut source
- Picture lock v6
- Embargo
- Holds until launch day
- Handed to
- Marketing lead — held
- Confidence
- 88%
- Held on
- Cue not licensed for promo use
As received
The brief, the cut it comes from and the embargo behind it — nothing here is inferred.
Evidence used
- Picture-lock v6 selects
- Cue sheet, promo terms
- Talent consent on file
Why it was held
The cue is licensed for the programme. A promo clears separately, not with it.
What the score decides
Confidence decides how much the lead re-checks, never whether an element is cleared.
Action
ReleaseRecutSend for clearanceValue
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 promo request
From the campaign calendar
Cut only from cleared material
Use the approved cut, the EPK stills, the cue sheet and the consents and approvals held against each element.
Turn one cut into a promo set
Cutdowns, stills, synopses and per-surface lengths are assembled from the cleared selects first, so the lead reviews a set instead of commissioning one.
Name the spoiler before it ships
A plot point, an unannounced date, a cue with no promo licence or a likeness with nothing on file is raised while the asset can still be recut.
Build an evidence trail
Retain the cut version, the stills and cues used, the consents read, the approval given, the lead's recut and where the asset ran — on both paths.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
Media & post
- Avid MediaCentral
- Frame.io
- Select reels
- render queues
Assets & EPK
- Iconik
- Dalet
- EPK stills
- Artwork libraries
- brand templates
Rights & consents
- Cue sheets
- promo music licences
- Talent consents
- billing and credit terms
Agent
Marketing-content generation
- Assembles cleared material
- Marks spoilers and gaps
- Holds for the lead
Publishing & scheduling
- Social publishing
- owned channels
- Trailer distribution
- embargo calendars
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 release
Each control wraps the one inside it. An asset clears every layer before a lead can release it, and creating a performer's likeness or voice sits outside all six.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Decide what material may be used at all
- L3
- Decides what may be shown yet
- L4 – L5
- Hold the release to a person and log it
- L6
- Pulls automation back when signals degrade
How Nestack evaluates it
Evaluate the asset and what it was cut from.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the promo asset the marketing lead opens
- E1Final-output evaluationDid the elements used trace back to a cleared source?
- E2Step-level evaluationDid it read the current cut, cue sheet, consents and embargo date?
- E3Tool evaluationDid it pull the right cut version and write to the right campaign?
- E4Spoiler and embargo recallWere plot points and pre-embargo details caught before release?
- E5Slice evaluationHow does the assembly hold across titles, surfaces and territories?
- E6Business outcomeHow much was recut before release, and what was pulled after it?
Floor — the assets recut before release or pulled after it
Failure modes
Where each failure originates in the agent
Seven failure modes plotted against the five stages of the agent lifecycle. The ones that matter most are the ones a takedown cannot undo.
Agent lifecycle
Direction of processing →
01 · Material & rights
2 modes
- PO-01
Promo cue not licensed
A programme licence is taken as covering it.
- PO-02
Generated element, no consent
A supplied synthetic shot has no consent record.
Stage gathers
The cleared cut, stills, cue sheet and consents
02 · Assembly
2 modes
- PO-03
Spoiler in the cutdown
The shorter cut leaves the twist in frame.
- PO-04
Credited cast cropped out
Key art drops a performer the contract keeps in.
Stage builds
The cutdown, the key art and the copy draft
03 · Rules check
1 mode
- PO-05
Promo exceeds its rating
A cutdown shows what the band it carries does not allow.
Stage checks
Spoilers, embargo dates, rating and territory
04 · Handover / release
1 mode
- PO-06
Embargo missed at release
The asset is queued ahead of the date it may run.
Stage presents
The asset the marketing lead reads and releases
05 · Change / Version
1 mode
- PO-07
Silent rule regression
A template change drops the rating or disclosure line.
Stage tracks
Model, prompt, rule and rights-record changes
- Sev-1 · a consent, licence or credit term is broken
- Sev-2 · the promo shows what it should not
- Sev-3 · assets drift and more is recut
Affected slices
The titles with the least material carry the recuts
Titles with no locked cut, thin archive paperwork and first-look assets built before an embargo lifts carry most of what a marketing lead recuts and most of what is pulled after release. Nestack reports performance by slice, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Unscripted and live formats | 4.5% | 2.9× | Review | |
| Archive and library titles | 3.6% | 2.3× | Review | |
| Pre-embargo first-look assets | 2.8% | 1.8× | Watch | |
| Returning-series episodic promos | 1.1% | 0.7× | Normal |
Bar: recut-and-pull rate lift vs. returning-series promos · scale 0–4.0× · tick marks the 2.0× threshold
2 of 4 slices over threshold
Evidence-linked improvement
An asset that has gone out is already copied
A pulled trailer stays up somewhere. What the cycle can change is the next asset, not the one a takedown is chasing.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Recuts, rights holds or a pulled asset cluster on a title or a surface.
- 02
Diagnose
Traced to the cut version read, a rights record, the spoiler rule, the embargo date or the assembly.
- 03
Improve
The record, rule or template is corrected with your marketing lead and business affairs, and version-linked.
- 04
Verify
Re-run against held-out briefs from that cohort, including the assets that were pulled.
- 05
Learn
The pulled asset is kept as a worked example and the rule it broke enters the promo checklist.
Learn → Detect
The return edge. Consents, promo licences and embargo dates are re-read each cycle — an element cleared for the last campaign may not be cleared for this one.
Typical build scope
Twelve workstreams across six weeks
The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, source and rights records, assembly, evaluation, integration, then supervised campaigns and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Workflow discovery and automation-boundary definition.Week 1
- 02Cut, select-reel and EPK source assessment.Week 1, Week 2
- 03Talent consent, likeness and credit terms.Week 1, Week 2
- 04Music cue and archive-clip promo licences.Week 2
- 05Spoiler, embargo and release-date rules.Week 2, Week 3
- 06Cutdown, key-art and synopsis assembly.Week 3
- 07Rating and territory rule sets.Week 3, Week 4
- 08Marketing-lead review and release workflow.Week 4
- 09Evaluation suite, slices and regression campaigns.Week 4, Week 5
- 10Territory versions and language checks.Week 5
- 11Publishing, scheduling and channel integration.Week 5, Week 6
- 12Observability, deployment 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 title, one surface | ProductionProduction asset integration | AdvancedMulti-title / multi-market |
|---|---|---|---|
| Introduced at Pilot | |||
| Cutdown, key-art and synopsis drafts | ✓ | ✓ | ✓ |
| Elements bound to their cleared source | ✓ | ✓ | ✓ |
| No generated or altered performer likeness | ✓ | ✓ | ✓ |
| Spoiler, embargo and rating holds | ✓ | ✓ | ✓ |
| Named marketing lead releases the asset | ✓ | ✓ | ✓ |
| Baseline evaluation | ✓ | ✓ | ✓ |
| Rights-record and cue-sheet reading | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Media, MAM and EPK integration | — | ✓ | ✓ |
| Review workflow and observability | — | ✓ | ✓ |
| Title libraries and archive licence sources | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Multi-market versions and enterprise controls | — | — | ✓ |
| Build price | From $5,000 | From $8,000 | Custom quote |
- Final build price
- Confirmed after discovery based on titles and surfaces in scope, media and rights-system integrations, asset volume, territory and language coverage, 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
01The approved cut and the select reels you promote from
→Cut, select-reel and EPK source assessment
Week 102Talent consents, billing and credit obligations by title
→Talent consent, likeness and credit terms
Week 103Cue sheets, and what your licences allow in a promo
→Music cue and archive-clip promo licences
Week 204Your spoiler rules, embargo dates and announcement plan
→Spoiler, embargo and release-date rules
Week 205The ratings and territories each promo has to sit inside
→Rating and territory rule sets
Week 306Assets you have pulled or recut, and the reason given
→Evaluation suite, slices and regression briefs
Week 407Named marketing leads, and who clears a right
→Review and release workflow, then supervised campaigns
Weeks 5–6
Nothing else is required
Deployment, documentation and Agent Care handover are ours.
Delivery timeline
Four phases across six weeks
Phases are drawn over the weeks they actually occupy. Week 5 carries both the territory checks and the first assets a marketing lead releases.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Titles and surfaces in scope; who clears a right and who releases
W2Cut, EPK, cue-sheet and consent records wired in
W3Cutdown, key-art and synopsis assembly, and the spoiler rules
W4Rating rules, review workflow, evaluation suite and slices
W5Channel integration and the first assets released under sign-off
W6Campaigns running under your marketing lead, then Agent Care starts
- Reading the band
- Consents, cue sheets and embargo rules are in place by the end of week 2, before the first frame is assembled in week 3.
- At the end of W6
- Assets have gone out under a named marketing lead, with the consent, licence and approval behind the elements used on file, then Agent Care takes over monitoring.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Media AI agent
Build a promo agent around your cut and your clearances.
Show us a campaign you are running now, the cut and stills it draws on, and where your consents and cue sheets live. We'll assemble one promo set from your own cleared material and show you the elements it refused to touch.
- Typical delivery
- 4–6 weeks
- Deployment
- Your media and MAM stack
- Release
- Named marketing lead, always
- After launch
- Agent Care available
Nestack Agents · Marketing-content generation
AGT-ME-05 · Agent Care available after launch