KevinDunnell

Max-Extraction Minimalist | PhD candidate at MIT @medialab, Human-AI Interaction, latent space exploration |

[Mehul Agarwal @meh_agarwal]

Hollywood isn't dead. It's evolving. And we're leading that evolution.
We just shipped Koyal v2.5: The best Agentic AI filmmaking platform. It goes from your script or music to full video with consistent characters, settings & storylines.

Large Production houses, music labels & ad agencies use @KoyalAI for storyboarding & pre-viz.

Smaller studios use it for making content they never had the time, budget or resources to create.

Write a scene. Direct the camera. Build a world. @KoyalAI brings it to life.

Here's what's new:
- Go from script to video:
Write a scene with dialogue, pacing, emotion. Koyal fully realizes it. Using the best voice models on the planet.

- Direct the camera:
Push in. Pull back. Drift through a scene. Real 3D camera blocking. Shape the shot the way you see it in your head.

- Build a world that stays:Lock in a location and return to it. Your scenes stay consistent from the first frame to the last.

Plus: dialogue, SFX, annotations for editing, and more control over every shot.

Under the hood, we benchmark 40+ models weekly with real humans so Koyal always uses the best available (more on that next week!)

Our agents pick the best models for each scene in run-time so you focus on the story, we handle the complexity.

You don't need to prompt a film, you can direct it.

With Koyal, we're looking to replace the camera, not the filmmaker behind it.

Try it now at beta [dot] koyal [dot] ai

DM me for credits!

[KevinDunnell @KevinDunnell]

@Benioff @adcock_brett Will we still call it a ‘fulfillment center’ if robots free humans for more fulfilling tasks?

[KevinDunnell @KevinDunnell]

Can tech help us fight back against the attention economy?

At MIT’s HARD MODE Hackathon we built Feral Focus — an ambient AI attention companion that lives on your desk.

🏆 Won the Reflect track.

Built with @MateoFerfer007 @safiware @realxdd44

Big thanks to @cyrusclarke for organizing such a hardcore event 🔥⚙️

@AHA_medialab @hardmodemit #HARDMODE

[designboom @designboom]

modular momo habitat by @medialab self-assembles on the moon https://t.co/MAA8Lcnefl

[designboom @designboom]

modular momo habitat by @medialab self-assembles on the moon https://t.co/MAA8Lcnefl

[KevinDunnell @KevinDunnell]

@enjalot 🖐️

[KevinDunnell @KevinDunnell]

@BrianRoemmele Seems like a waste - printer could be programmed to print interesting, structurally stable organic forms, yet they choose a form that could be achieved with stacking cinder blocks 😂

[KevinDunnell @KevinDunnell]

@ErikVoorhees ✋

[KevinDunnell @KevinDunnell]

Sweet! We’ve been exploring organization + synthesis of patents w LLMs in context of USPTO db as well as other org’s datasets - https://t.co/qOYDHuicVl publicly showcases a 10k subset of the 7M patents we processed. If there’s an opportunity to collab on visuals/interactions and/or get your feedback I’d love to chat!

[KevinDunnell @KevinDunnell]

Yes, Grok’s performance is lagging other frontier LLM’s, but it could be well positioned as a profitable consumer LLM (and maybe help us better understand our universe)

- Grok is likely trained on X’s own platform data (hedge against legal fallout from data scraping)

- Grok has access to a larger stream of information than broadcast news (other LLM / MoE will be restricted to non-X search results)

- X is openly pursuing a payment platform and micro payments can enable pay-per use AI APIs (Subscription models don’t seem to scale here as user activity increases - see Microsoft’s Co-Pilot)

#AI #GenAI @xai #Grok @OpenAI #ChatGPT

[Brian Roemmele @BrianRoemmele]

🚨🚨GROK NEWS FLASH 🚨🚨

Grok will have a native payments system built-in and likely be released early next year. Grok tribe members will be able to monetize their SuperPrompts and isolated datasets, including X postings.

This will raise the stakes of cloud AI platforms.

[KevinDunnell @KevinDunnell]

@OfficialLoganK @OpenAI Will GPT-4V API offer image embeddings? Will it be similar 1536-dimension space as Ada text embeddings? Asking for an app… (https://t.co/MS35P2rNak)

[KevinDunnell @KevinDunnell]

@willdepue Awesome! With https://t.co/MS35P2rNak we’ve embedded all 7M+ USPTO patent descriptions (10k sample via link) Beyond search + exploration of existing content, embeddings-landmarks in high dim. space of working solutions-will help guide meaningful synthesis. Would love to collab!

[KevinDunnell @KevinDunnell]

@katiedunnell @PicturesFoIder 😂

[KevinDunnell @KevinDunnell]

@dylanleclair S’mordinals

[KevinDunnell @KevinDunnell]

All hidden states across all layers of #LLaMa-7B, concatenated and mapped to bitmap image for each token...

The first hit of #SnowCrash is free

[KevinDunnell @KevinDunnell]

You must feed the machine just enough to escape it without it consuming you in the process

[KevinDunnell @KevinDunnell]

@andrewchen Agreed! Been hacking on a cleaner UX for this use case - https://t.co/6tMrUplrsV