How to Onboard an AI Production Partner in Two Weeks: The Commissioning Side Checklist

Contract signed on a Friday. First batch expected in six weeks. Nobody on the commissioning side has been told they own anything.

Three weeks later the production partner is waiting on a script lock, a character reference decision and an answer about which territory the series is for. None of those are production problems. They are onboarding problems, and they consume schedule that was allocated to making episodes. The partner absorbs the first week of silence, flags the second and re-baselines the delivery date in the third. By the time the first batch arrives, the series is late for reasons that had nothing to do with production capability.

Onboarding is a commissioning side responsibility and it fits comfortably into two weeks if someone runs it. What follows is the sequence that works, organised by what has to be true at the end of each block rather than by activity.

1. What the Two Week Window Is Actually For

Onboarding is not relationship building. It is the transfer of every decision the production partner needs in order to work without asking you, plus the establishment of a rhythm for the decisions that cannot be transferred in advance. Measured that way, it either succeeded or it did not, and the test is simple: how many blocking questions does the partner have to raise in the first production batch.

The cost of skipping it is not visible at signature. It appears as latency. Every unanswered question becomes a pause, and pauses on a 70 episode series compound because generation, review and edit are sequenced. A question that takes you four days to answer does not cost four days of the partner calendar. It costs four days plus whatever it takes to restart the workstream that stopped.

Two weeks is the right length because the window before production start is the only period where decisions are cheap. A character design decision made in week one costs a conversation. The same decision made in batch three costs the episodes already generated against the old design. Nothing about onboarding is urgent in the ordinary sense, which is exactly why it slips, and why it needs a date and an owner like any other deliverable.

2. Days One to Three: Name the Owners on Your Side

Before any document moves, name people. Vague ownership is the single largest source of onboarding delay, because a question addressed to a company is a question addressed to nobody.

Four roles need names. A commissioning lead who owns scope, budget and the final creative call. A reviewer of record who consolidates all notes into one voice and holds the review calendar. A rights and compliance contact who can answer clearance questions without escalating. A technical delivery contact who owns specifications, file handling and the acceptance test. On a small team one person can hold two of these. No role can be held by nobody.

Write down decision rights explicitly, including what the reviewer of record can approve alone and what must go to the commissioning lead. The reason this matters is that most onboarding delay is not disagreement. It is a reasonable person waiting for permission they already had. In the wider production world the boundaries between producer and director authority are long established, and the role definitions used across producing and directing work are a useful reference point for a team assigning them for the first time in an AI native context.

Give the partner one named escalation path with a response commitment attached. Twenty four hours for a blocking question, three working days for a creative decision, and a standing weekly slot for everything else. The commitment matters more than the numbers.

3. Days One to Three: Assemble the Document Pack

Everything the partner needs should arrive as one pack rather than as a fortnight of attachments. Assembling it is tedious and it is the highest return work in the entire onboarding window.

The pack contains the approved brief, the series bible or outline, the locked script or the script delivery schedule, the character reference set with approval status marked, the visual and tonal references, the technical delivery specification, the territory and language plan, the release cadence if known, and the list of names with decision rights from the previous step.

Mark the status of every item honestly. An outline described as locked that is still with a writer is worse than an outline described as draft, because the partner will build against it. Three labels are enough: locked, draft, and not yet started with a date. A pack that is forty percent locked and labelled accurately is workable. A pack that is ninety percent locked and labelled optimistically produces rework.

Send the specification even if it feels premature. Resolution, frame rate, aspect ratio, codec, loudness target, subtitle format, file naming convention, delivery destination. These are the cheapest items to agree and the most expensive to discover at delivery, when a technically excellent series fails a platform ingest check on a loudness standard nobody mentioned.

4. Days Four to Seven: Lock the Creative Inputs

By the end of week one the partner should be able to produce a shot without asking a question about what it is meant to contain.

Character references first, because everything downstream depends on them. Each principal character needs an approved reference set, a written description of what must stay constant, and an explicit note on what is allowed to vary. Wardrobe changes, ageing, injury states and any transformation the arc requires should be identified now rather than discovered at the episode where they occur.

Then the tonal and visual standard. One page, with references, describing what the series should look and feel like and naming the two or three things that would be wrong. Naming what is wrong is more useful than naming what is right, because it is falsifiable. A note saying the series should not look like a glossy commercial gives a generation operator something to check against.

Then the brief itself, reviewed once more against the question of whether it is executable by someone who was not in the room. That is the actual standard for a brief. If a competent person outside your organisation cannot produce the intended result from the document alone, the document is incomplete, and the gap will be filled by guesswork that you will pay to correct later.

Finally, the script position. Either the script is locked, or there is a written schedule for when each block locks, with the understanding that production sequencing depends on it. Rolling script delivery is workable. Rolling script delivery that nobody has scheduled is not.

5. Days Four to Seven: Agree the Operating Rhythm

The rhythm is what carries the production once onboarding ends, and it is almost always set by default rather than by decision.

Agree batch size. Ten episode batches are the workable default for a 70 episode series, because they surface systemic problems early and remain small enough to review properly. Agree the review window in working days and make it binding on your side, because commissioner review latency is the most common cause of a slipped delivery date and the easiest to prevent.

Agree the note format and the single channel it travels through. Shot identifier, observed problem, reference breached, required outcome, priority. One note per issue. One consolidated list per batch from the reviewer of record. Notes arriving in three places from four people will cost a production cycle every batch.

Agree what reporting you will receive and at what interval. Shot completion against plan, correction rate by cause, open blockers, and the next decision you owe. That last item is the one most commissioners have never asked for and the one that most reliably prevents silent delay.

Agree the standing meeting and keep it short. Thirty minutes weekly, with an agenda that is the report plus decisions. Meetings that exist to discover status are a symptom of missing reporting.

6. Days Eight to Ten: Rights, Clearance and Compliance

This block is where onboarding most often stalls, because it involves people who were not part of the commissioning conversation.

Establish the chain of rights in writing. Underlying IP, adaptation rights, script ownership, music, and the ownership position on delivered assets including the reference sets and any trained character assets. The last item is frequently left undefined and it determines whether a sequel can be produced at all, so it belongs in the onboarding pack rather than in a renewal negotiation eighteen months later.

Settle the performer and voice position early. Whether any human performance, likeness or voice is involved, what consents exist, and what the disclosure obligations are in each release territory. The frameworks here have moved quickly, and the contract and industry resources published by SAG-AFTRA are the practical reference point for a commissioner working out which agreements apply to a given production shape.

Confirm platform and territory compliance requirements. Age rating, content restrictions by market, disclosure requirements around synthetic content, and any platform specific acquisition criteria. A series that cannot be delivered into its intended market is a total loss, and the check takes a day.

Name the single person who answers clearance questions during production. Without that name, every clearance question becomes an escalation and every escalation is a pause.

7. Days Eleven to Fourteen: The Calibration Batch

Do not go straight to volume. The last block of onboarding is a small calibration batch whose purpose is to test the whole pipeline rather than to produce usable episodes.

Commission a short sample: a handful of shots covering one principal character, one dialogue scene, one hook and one technically difficult moment such as a night exterior or a physical action. Run it through the complete path, including your review, your note format, your consolidation process and the technical acceptance test. The point is to break the process while it is cheap.

Review what the calibration batch teaches about your side, not just theirs. How long did review actually take against the agreed window. Did notes arrive in the agreed format. Did the acceptance test catch what it was meant to catch. Did anyone have to ask a question that the document pack should have answered. Every one of those findings is worth more than the footage.

Approve the calibration batch formally, in writing, and state what is now locked. Character references, tonal standard, technical specification, note process. Locked means changes from here are change requests rather than corrections, which is the boundary that protects both the budget and the relationship.

Then start production with the delivery schedule re-baselined from the calibration approval date rather than from contract signature. A date set at signature and never revisited is a date that will be missed for reasons nobody recorded.

8. What Slips and How to Stop It

Four things slip, predictably, and each has a cheap countermeasure.

Script lock slips, because writing is not a two week activity. The countermeasure is a written block schedule rather than a single lock date, agreed with the partner so production sequencing can be built around it.

Character approval slips, because it requires taste and taste requires a meeting. The countermeasure is a named approver with authority to decide alone, and a default that applies if no decision arrives by the date.

Legal review slips, because it sits outside the commissioning team. The countermeasure is to start it on day one rather than day eight, since it is the only block in the sequence with a dependency you do not control.

Reviewer availability slips, because nobody costed review time. The countermeasure is to put the review windows in the reviewer calendar during onboarding, as blocked time, before production starts.

A two week onboarding that runs to three weeks because legal took longer is a success. A six week onboarding that nobody was running is the version that shows up as a late series.

Axis AI Studios Perspective

Axis AI Studios runs onboarding as a defined block with a document pack, named owners on both sides, an operating rhythm agreed in advance and a calibration batch before volume production begins. The delivery schedule is baselined from calibration approval rather than from signature, because a schedule that ignores the decisions still outstanding on the commissioning side is a schedule that will move.

We are an AI native vertical drama production studio working with platforms, media companies, brands and IP holders across commissioning, production and delivery. Current production partners include Den Tolmor and Good Fight Production LLC and HolyWater. Commissioners running their first AI native series usually find that the hard part is not evaluating the partner. It is discovering how many decisions their own organisation has not yet made, which is precisely what a structured onboarding window is for.

If you have selected a production partner and want the onboarding sequence run properly before the first batch, write to business@axisaistudios.com.

FAQ

Can onboarding realistically be done in two weeks on a first commission?

Yes, with one exception. Everything in the sequence is within the commissioning team control except legal and rights review, which depends on a function that has its own queue. Start that block on day one rather than in week two and the remainder fits comfortably into fourteen days. Where two weeks genuinely fails is when the script is not written, in which case the correct answer is a block delivery schedule rather than a compressed onboarding.

What is the single most important item in the onboarding pack?

The approved character reference set, with a written statement of what must stay constant and what is allowed to vary. Everything generated downstream is measured against it, and a reference set that is changed after production begins invalidates the episodes already produced against the previous version. Scripts can arrive in blocks. Character references cannot arrive late without cost.

Should a commissioner pay for a calibration batch separately?

Usually it sits inside the base fee as a defined production stage, and it is worth insisting that it exists whether or not it is priced separately. The value is not the footage. It is that the reference set, the note process, the review window and the technical acceptance test are all tested while the cost of a failure is a handful of shots rather than a batch of episodes.

Further Reading

For the evaluation work that precedes onboarding, the production partner due diligence checklist covers the six domains worth assessing before a partner is selected.

For building the central document in the onboarding pack, the production brief guide covers what a brief has to contain to be executable without clarification.

For what happens after onboarding ends and the common failures in the period that follows, the first ninety days of commissioning covers the decisions that most often go wrong early and how to prevent them.