@Thomas Squires , founder of Yuumeh, came on live today. He showed us the platform for over an hour. Generate for free, test run the platform: it's for AIFD members, free or Pro. DM me "Yuumeh" with your profile link and I'll send you the sign-up link. No profile yet? Make your free one in 2 minutes: https://aifilmmakersdirectory.com/add-new-business And we tested Arabic live. Part of it worked. Part of it crashed on camera. More on that below. Watch the replay 👇 What stood out: 1. Plug in your own AI. You can link Claude (and other tools) to Yuumeh through connectors. Same workspace, long multi-step projects, and you can run it on the Claude plan you already pay for. That's how I made the shots for my "Free or Pro" video. 2. Consistency on autopilot. It builds character sheets, location sheets and storyboards by itself, before it makes a single shot. 3. Voice first. It makes the dialogue and sound before the video, so lips and audio line up better. 4. A timeline editor built in. Snap clips together, trim them, and have it generate a new shot to fill a gap. Thomas made a full launch trailer inside Yuumeh: edit, titles, lighting, sound mix. No other editor. 5. Swap tools. Restyle, expand, motion transfer, lip-sync, and character swap (drop your character into an existing video). 6. Sound. Dialogue, music, ambient soundscapes, with ElevenLabs inside. 7. You stay the director. Thomas said it best: the platform handles the technical work, you keep the creative call. "You direct. Yuumeh produces." Coming soon, per Thomas: - A library of cinematic references, to keep skin, lighting and camera work out of AI-artifact land - Several agents working at once, built for micro-dramas (whole episodes in parallel) - A separate spot for ambient sound, apart from music The Arabic test. Big welcome to Aya, who jumped in live and asked the questions we all needed asked. Arabic text and Arabic voice came out fine. The lip-sync test with a regional accent fell flat: wrong line, wrong accent. Thomas thinks the audio got turned into text before it reached the video model. He's testing Arabic lip-sync and will post the result here.

At 12:00 noon ET, Thomas Squires joins me to introduce us to his platform, Yuumeh: "You direct. Yuumeh produces." Bring your questions, they're welcome. And Thomas is giving free credits so you can try it out (the first 20 to sign up get 1,000 credits). Don't miss it. Come spend an hour with us: https://meet.google.com/edu-vrqe-fvs

Big one coming Monday. @Thomas Squires , the founder of Yuumeh (yuumeh.com), is joining us on AIFCA for a live recording session. He's going to demo the platform, and you get to ask him your questions. Why I'm pumped about this one: I've been testing Yuumeh and it's very good. Connecting it to Claude Code was a breeze. I let Claude Code do the heavy lifting to generate the shots I used in my "Free or Pro" video, the directory's new 3-minute video. And members get free Yuumeh credits to try it out [how many credits, once Thomas confirms]. Monday, Nov 9,12:00 PM ET. Come live and bring your questions: [link to the event] Can't make it? It's recorded, so you can catch it after.

Create videos up to 30 seconds at the lowest price available, including 9:16 in our V5.0 launch Beta for AI Vertical "Microdramas". Request free: Reply, DM, email [email protected], or visit http://createsagas.com for our contact form. Followers of our new brand @CreateSagas get priority access so follow now! Restrictions apply. No celebrity likenesses or offensive/X-rated content allowed. Limited time and credits given during Beta.

Remember the Claude Code + Higgsfield class I recorded with Sean? It ran 1 hour 47 minutes. Raw. Lots of good stuff in there... and lots of us fumbling with logins, screen-share glitches, and me hunting for a tab I'd closed. Normally that's an afternoon of scrubbing through a timeline. Instead I told Claude Code, in plain English... from my phone. I wasn't even at my desk. I gave every instruction remotely, and Claude Code did all the work on my desktop computer (transcribing, scanning the video, building the edit). Here's what I told it: - transcribe it and make a rough cut in Cutroom (the editor Claude built for me, live, in that same class) - cut the dead spaces, but leave 15 frames of handle on each side of every cut, don't cut tight on the words - don't cut silent moments where we're navigating on screen, people need to SEE how it's done - look at the big picture and cut the stuff we wasted time on that turned out to be easy in hindsight (permissions, login loops, etc.) It transcribed the whole thing, scanned the video frame by frame to see when the screen was actually moving, read the full conversation, and came back with this report. Here it is, just reformatted for Skool: --- The rough cut is ready in Cutroom: "Claude Code + Higgsfield Class w/ Sean (Rough Cut)". It runs 75.7 minutes, down from 107, in 57 clips. I haven't watched it back or exported it. What got cut: - The 12 detours, about 30 minutes in total: 00:00-13:02 Pre-show chat, slow Higgsfield login, closing tabs and apps. Opens on "Hi everybody, we are today with Sean Rubin" 15:24-16:38 The "I don't have DaVinci" / CapCut detour before setup 17:19-25:14 Failed auth attempts: /mcp, searching Connectors in the desktop app, Apollo picked by mistake 26:05-28:36 Link-pasting back and forth before the fix 49:54-50:13 Voice-input test that didn't come through 50:17-50:25 "I was distracted, tell me again" 58:45-59:43 Screen-share glitch; video not visible, re-sharing 1:00:12-1:00:36 Stopping and restarting the screen share