Lesson 60.1 gave you the pipeline and the previs packet: explore cheaply, commit deliberately. This lesson answers the next question — who does the work when exploration gets cheap? The honest answer is smaller teams with clearer ownership, not one person doing everything forever.

By the end of this lesson, you can say what gets cheaper, what never disappears, which emerging roles cover the new work, and where a named human stays responsible.

What actually gets cheaper

Generative media lowers the cost of five things:

1. Visual exploration. Dozens of directions, references, and variations before committing to a design. 2. Rough assets. Stand-ins, backgrounds, crowd fill, texture studies, and layout drafts that used to take days. 3. Localization drafts. First-pass translated dialogue, timing guides, and caption structures for review by a fluent speaker. 4. Versioning. The same scene reframed for a trailer, a poster direction, a vertical cutdown, or a pitch deck without rebuilding from zero. 5. Post-production experiments. Trying three grades, three sound beds, or three edit rhythms on the same material in an afternoon.

Cheaper does not mean free, and it does not mean finished. Each of these still needs selection, correction, and approval. What changes is how many options a small team can consider before spending real production money.

What does not disappear

Every project that reaches an audience still needs these, regardless of tooling:

• Story. What happens, why it matters, and why this order. No model decides what your film is about.
• Performance. What an actor, voice performer, or presenter brings that cannot be prompted: timing, restraint, surprise, lived presence.
• Taste. Choosing among ten good options. Tools generate options; taste eliminates nine.
• Directing. Holding one coherent intention across hundreds of small decisions.
• Production management. Schedule, budget, locations, safety, communication. The least glamorous work and the first thing that sinks a shoot when missing.
• Rights and clearance. Permission for every person, voice, location, song, brand, and artwork the audience can recognize.
• Craft. Cinematography, editing, color, sound, music, design — the practiced hands that make the final 10 percent feel inevitable.
• Audience knowledge. Who this is for, where they watch, what they expect, what earns their trust.
• Editorial judgment. What to cut, what to keep, what to verify, and what to take responsibility for publishing.

If someone tells you AI removes all of these, they are selling software, not describing production.

The emerging roles, without hype

Smaller teams still need the work divided. These practical roles are appearing on real productions — titles vary, but the jobs are consistent:

None of these is "the AI person who does everything." Each one pairs a tool capability with a human responsibility. On a two-person short, one person may cover three roles — but the responsibilities still need names next to them, or they quietly go undone.

The small-team advantage versus the one-person-forever fantasy

A small team with these roles has a genuine advantage: it can test more ideas, move faster between previs and production, and keep creative control close to the people making decisions. Two careful people with a clear packet can now do exploration work that once required a department.

The fantasy version says one person does every job — writing, directing, performing, shooting, editing, scoring, marketing — forever, because the tools exist. That collapses for predictable reasons:

• Review blindness. You cannot catch your own story hole, legal risk, and color problem in the same tired pass at midnight.
• No coverage. When one person stalls, the production stalls. There is no second judgment to keep it moving.
• Ownership fog. If everything is "AI-assisted," nobody can answer who approved the face, the claim, or the final cut — until an audience or a rights holder asks.

The rule is simple: a small team is a strength; an unreviewed solo chain with no named owners is a liability. Deliberate review and clear ownership are the difference.

Build it: map your workflow with owners

Take a video or film workflow you know — a short, a lesson video, a promo — and map it in three passes:

STEP | HUMAN-LED | AI-ASSISTED EXPLORATION | REQUIRES PERMISSION/APPROVAL | OWNER

Worked example for a 3-minute explainer video:

Mark every step with at least one label, and give every approval step a named owner — a person, not "the team" or "AI." Steps involving a real person's face, voice, or property, any factual claim, and the final publish decision always need a named approver.

Check your understanding

1. Name two things that get cheaper and two things that never disappear, and explain the difference in one sentence each. 2. Pick two roles from the table and describe what breaks on a production if nobody fills them. 3. Why does a three-person team with named owners outperform a solo creator with the same tools on anything meant for a real audience?

Common failure mode: a map where every step is "AI-assisted" and no step has an owner. That is not leverage; it is an unreviewed pile. Re-mark the map until at least story, performance, rights, and final approval are human-led with names attached.

The finish line

You are done when you hold a team and role map: every workflow step marked human-led, AI-assisted exploration, or requires permission/approval — with a named human owner on each approval. That map shows where AI creates leverage and where responsibility lives.

Next, Lesson 60.3 takes on the hardest part of that map: likeness, permission, and audience trust — the boundary that decides whether your production survives contact with the real world.