People DM me this question every week: "What tools are you using?" The honest answer is that the stack changes every six months because the AI space moves that fast, but the roles stay the same. Here's the current 2026 stack with every tool I actually open during a typical project, why I use each one, and what I'd swap if it disappeared tomorrow.

I'm grouping them by job, not alphabetically, because what matters is what each tool does in the pipeline.

1. Writing & ideation: ChatGPT and Claude

Both. Not one or the other. They're good at different things.

ChatGPT is faster and looser. I use it for brainstorming, hook generation, viral title testing, and quick rewrites. When I need 20 versions of an opening line so I can pick the best one, ChatGPT.

Claude is structured and patient. I use it for long scripts, character bible writing, and any task where I need it to remember 15 pages of context without drifting. When I'm building a 7-part historical series and need consistent voice across episodes, Claude.

Both have free tiers that are usable, but if you're producing weekly you'll want at least one paid subscription. ChatGPT Plus runs $20/mo, Claude Pro runs $20/mo. I pay for both because they save me 10x their cost in time.

2. Image generation: MidJourney + Nano Banana + Flux

Three image models because they're each best at different things.

MidJourney v7 is still king for stylized, atmospheric, painterly imagery. If I want my scene to look like a 1990s book cover or a Studio-style anime frame, MidJourney handles it natively without complex prompting. The new --cref character reference parameter solved a lot of the consistency problem.

Nano Banana Pro (Google's Gemini 3 image model) is the new daily driver for multi-element compositions. It accepts up to 14 reference images, handles text rendering at near-perfect accuracy, and produces 4K outputs cheaply. For thumbnails and complex scenes, it's now my first choice.

Flux (the local/open-source workflow via ComfyUI) is what I use when I need fine control. Custom LoRAs, exact composition, ControlNet posing. Slower setup, but the only option when you need surgical precision.

3. Video generation: Runway, Kling, and Seedance

Image-to-video is the part of the pipeline that still requires testing. Different models nail different motion types.

I run a shot through whichever model fits the motion, not whichever I "prefer." The cost per shot ($0.50 to $2.00 depending on model and length) is too low to be precious about it.

4. Voice & audio: ElevenLabs

For narration, character voices, and any spoken-word audio, ElevenLabs is still the only realistic option. The voice cloning is now nearly indistinguishable from real speech, and the multilingual support means you can dub a video into 30+ languages from one English script.

For background music, I use a mix of Suno for original tracks I want to own and Epidemic Sound for licensed tracks I just need to slot in fast. Sound effects come from Soundly and free Freesound libraries.

If your audio is bad, your video is bad. Most "AI animations look like AI animations" feedback is actually "the audio gives it away." Spend disproportionate time here.

5. Editing: CapCut or DaVinci Resolve

For short-form (vertical, under 90 seconds), CapCut Pro on desktop is fast, free, and gets out of your way. The auto-caption feature alone saves an hour per video.

For long-form (YouTube, anything over 3 minutes), DaVinci Resolve. It's free, professional-grade, has a real color grading suite, and exports cleaner audio than CapCut.

I do not use Premiere or After Effects for AI animation work. They're great tools, but the workflow is heavier than this content needs.

6. Asset management: Eagle

The unsexy tool I'd never give up. Eagle is a desktop app that organizes every reference image, every generated frame, every screenshot. After 18 months of AI animation work, I have ~40,000 images in there, all tagged, searchable, and organized into character libraries, style references, and shot databases.

The day I started using Eagle, my output speed doubled. Search "Cleopatra red room close-up" and I get every reference I've ever generated for that scene type, instantly.

The cheap version of this stack

If you're starting today and don't want to spend $200/mo on subscriptions, here's the minimum viable stack:

Total: $10/mo. Enough to publish 2-3 short videos per month while you're learning. Scale up only when the content starts paying for itself.

The expensive version

If you're producing weekly and treating this as a business:

Total: roughly $180-220/mo recurring. For a creator earning anything from this, it's the cost of doing business and pays for itself with the first ad campaign or sponsored video.

What I'd ignore

Tools I see hyped that I don't personally use:

◆ THE WORKFLOW

The full tool-by-tool walkthrough?

The academy has a 4-hour module showing every single tool above being used in real projects, with my exact prompts, settings, and templates. Plus a Slack-style community where students share new tool finds weekly.

Join the academy / $29.99 mo →
Tagged: Tools Workflow Runway MidJourney CapCut ElevenLabs