How do you plan and budget an AI-native production?

Plan an AI-native production around the finished scenes and the decisions needed to make them. Budget for development, design, generation, review, editorial, finishing and delivery, with explicit assumptions about what is already approved.

A running time alone is not an adequate specification. Two sequences of the same duration can require very different amounts of work. A controlled interview setup and an action scene with changing geography, recurring characters and complex interactions are different production problems.

Establish the version of the project

Identify whether the request is a development test, a pitch sequence, a production scene or a final delivery. Each has a different purpose. A development test can explore a question without resolving every element of the final film. A delivery sequence needs an agreed standard and complete supporting work.

Record the script or brief version used for the estimate. If the script changes later, the team needs a way to distinguish a creative development choice from work already included in the quote.

Ask which assets exist and what their approval status is. An image in a folder is not necessarily a production-ready character design. A performer reference may still need permissions. A location may look finished while its reverse angle remains unresolved.

Break down dependencies

List the sequences, then the shots and the elements they share. A returning character, a repeated room or a distinctive prop creates dependencies across the schedule. Work that settles one of those elements can make later shots easier; an unresolved decision can hold several scenes open.

Put those shared decisions before the work that depends on them. Avoid scheduling all generation as a single undifferentiated phase. A team may be developing one sequence while another enters editorial, provided the dependencies and review responsibilities are clear.

Name an owner for each approval. Waiting for a decision is part of elapsed production time even when no artist is actively generating material. Availability of the director, producer, editor and client reviewer belongs in the schedule.

Test a representative difficulty

A useful test should address a real risk in the commission. If the project relies on a recurring character interacting with props, a beautiful empty landscape does not prove that requirement. If the difficult issue is matching live action, test against a representative plate.

Define the decision the test must support. It may establish whether an approach is suitable, reveal the amount of finishing required or show that a scene should be filmed instead. Record the limitations as well as the successful frames.

A test can reduce uncertainty without justifying a claim that every later shot will take the same time. Complexity changes with action, framing, continuity and the notes received.

Build a cost structure

Separate development and shared asset work from sequence production. Then identify generation, artist time, supervision, editorial, finishing, sound, production management, storage and delivery costs. Add the rights and clearance work the project requires.

A quote should state what is included in those categories. For example, editorial may mean preparing material for an external editor or delivering a finished sequence. Those are different deliverables, even if both are described casually as “the edit”.

Keep allowances visible where information is incomplete. An allowance is a planning assumption, not proof that the work will fit inside it. Agree when an allowance will be replaced by a firmer scope and who can authorise the change.

Make revision rules practical

Define review stages and who consolidates notes. A team receiving conflicting instructions from several reviewers needs a decision process, not simply a promise of another revision round.

Explain which changes are refinements and which reopen an earlier approval. Changing a costume detail in one shot can be local work. Changing the approved character design may affect every scene where the character appears.

Allow time to review revisions in the sequence. A corrected individual shot can introduce a timing or continuity problem when placed back into the edit.

Plan delivery from the start

Confirm the required master, frame rate, aspect ratio, colour handling and sound deliverables. Ask whether the recipient needs textless elements, captions, separate components, supporting records or an additional version for another market.

Delivery also includes time for technical checks and correction. Do not schedule final review at the moment the master must leave. Keep a clear distinction between the team completing its work and the recipient accepting it.

Keep the schedule honest

• Identify what must be decided before production begins.
• Show review and waiting time alongside making time.
• Track dependencies across repeated assets and scenes.
• State which work can run in parallel and which cannot.
• Reserve a defined period for finishing and acceptance.

A usable budget explains the work and its assumptions. A usable schedule explains when that work can actually happen. Both should help the producer make choices before those choices become expensive revisions.

Keep the estimate connected to the approved scope as the project develops.