The Toolkit, Mapped

Every few weeks somebody sends me a list titled something like "500 AI tools every filmmaker needs." Nobody needs 500 tools. Most working filmmakers I know run four or five, and half of those quietly do the job of software they were already paying for. The trick isn't knowing every tool — it's knowing the categories, so when a new name lands in your feed you can file it in two seconds instead of losing an afternoon to a launch video.

So here's the map. Same pipeline you already work in, category by category, with the names you'll actually run into.

The categories, by pipeline stage

• Writing & development assistants — ChatGPT and Claude for brainstorming, coverage-style feedback, loglines, synopses, and producer paperwork. The most mature category and where most filmmakers should start.
• Image generation — Midjourney and its peers for concept art, storyboards, lookbooks, and pitch-deck frames. Previsualization, not final pixels.
• Video generation — Runway, Sora, Veo, and Flow producing moving footage from text or stills. The flashiest category; also the one where the demo-to-tool gap is widest.
• Voice & audio — ElevenLabs and similar for scratch voiceover, temp dialogue, and dubbing. Powerful, and wrapped in consent rules we cover in Chapter 4.
• Post-production assists — the AI already inside DaVinci Resolve and other editors: transcription, text-based editing, dialogue cleanup, upscaling, magic masks. The quiet workhorses.
• Marketing & distribution — caption generators, trailer assemblers, poster comps, social cutdowns. Small tools, real hours saved when you're your own marketing department.

Notice the shape of that list. The categories closest to the finished screen — video generation especially — are the least reliable. The categories buried in paperwork and post are the most reliable and the least advertised. That inversion is the single most useful thing to know about this market: the marketing volume of a category is roughly inverse to its current usefulness on a real production.

Free tiers, paid tiers, and the evaluation test

Almost everything on that map has a free tier, and the free tiers are genuinely good enough to learn on. That's not generosity — it's how these companies acquire you. Use it anyway. My rule: never pay for a category until a free tier has already saved you real hours in it. The moment a free tier becomes the bottleneck on actual work — you're hitting generation caps mid-storyboard, or the watermark is blocking a deliverable — that's your signal to pay, and not before. Prices shift constantly, so I won't quote numbers that will be stale by the time you read this; the pattern doesn't shift: free to learn, modest monthly to work, expensive tiers you probably don't need yet.

Before any tool earns a slot in your stack, run it through three questions. One: does it save real hours? Not "is the output cool" — did a task that took a day now take an hour, on your actual project? Two: can you export and own the output? If your storyboards live only inside someone's web app, behind their login, in their format, you don't have storyboards — you have a subscription with pictures in it. Real tools let you take your work and leave. Three: is it a tool or a demo with a pricing page? Plenty of products are one great generated clip on a landing page and a checkout button. If you can't find working filmmakers describing how it fits their pipeline — not influencers reacting to it — treat it as a demo until proven otherwise.

One more habit worth building now: keep your map organized by job, not by product name. Products merge, rebrand, and vanish — Runway today might be something else next year — but "I need boards for the pitch" and "I need this interview transcribed" are jobs that will outlive every logo on the list above. When the names churn, and they will, your map still works.

Next chapter we turn the map into a stack — how to actually choose your tools without bleeding money on subscriptions you touch twice a month.