Ex-Stuntman Gone AI

Joined April 2019
330 Photos and videos
Jun 12
Everyone talks about Fable and beleive me, i get it! But I can't believe how good and under radar Gemini is. We use gemini is most cases in our App and won't change it to anything else.
2
45
Jun 12
Also a big shoutout to @MiniMax_AI , underrated as f!
37
Jun 11
How it looks like: How it feels like:
26
May 27
1-Bit is mental
May 26
Today we’re releasing 1-bit and Ternary Bonsai Image 4B. A new family of image-generation models designed to run high-quality diffusion inference on local hardware: from laptops to phones.
4
2
177
May 21
Life's good. Let's test X Algo abit
3
1
2
177
May 13
Got stranded in Dubai for a layover night because of a delayed flight. Any builders want to grab a coffee? ☕️
4
223
Worst timing ever!
Pets. Now in Codex. Use /pet to wake your pet.
2
1
299
Apr 28
When Codex say this, you know you are cooked. Gpt5.5 is a beast of a model! Clears Opus 4.7 in many areas. But bro, don't you ever touch my UI again.
1
202
Apr 27
How it feels when claude code usuage limit resets

ALT Tai Lung Kung Fu Panda GIF

1
138
Apr 27
How using claude Design feels like as builder:
2
119
Apr 15
Exactly what I needed right now haha Gotta say the new Claude Code Desktop is very nice! I still used to run claude in cli and codex in app but now i might ditch cli more and more.
Apr 14
We've redesigned Claude Code on desktop. You can now run multiple Claude sessions side by side from one window, with a new sidebar to manage them all.
1
189
Behfar retweeted
Most brands can't afford a shortform video team. So we built one that runs on a single prompt. Script. Voice. Visuals. Edit. Done in minutes. This video? Made with VidGuy. Start to finish.
46
16
97
7,844
This is soooo expensive damn! Guess meant for enterprise only
Apr 6
🚨 sync-3 from @synclabs is now live on fal! 🎬 The next evolution in lipsync 🌍 Multilingual & multi-character support 🎯 Native visual intelligence, built for professional-grade video
186
Hot Take: I'm following open Source very carefully and I am super excited about it and I know it needs to win! 🫡 But as a founder at this stage, who needs to just focus on building, marketing and distribution. There is no time for "optimising" my token cost.
1
2
126
Cursor 3 is basically codex app and i love it!
1
6
498
Quick story on this: While building our AI companion Platform, one of the biggest focuses were Video Calls. Since Majority of our clients were game Studios, we were lucky to be able to collaborate with talented artists providing Rigged 3d Models for custom Video Call experiences. This was ver viable for client based work, but not a self service platform. We did in parallel, we did many experiments with Stitching opensource models , streaming content and caching and looping video generations to mimic a video call experience. We were not funded and still are not so we couldnt really afford to burn through tokens, we tried to be very "scrooch" with our approach. We saw some early results with a closed group and the feedback and the impression was always positive, there was something magical about not only hearing your favourite character, but also seeing them while having a conversation. However we quickly find out that it's just too expensive for a consumer use case but i do believe there is a huge opportunity here and as the models get better, more squashed to cut down cost (shoutout to Pruna for example) , this is going to be a very exciting field for Character IP holders
The cost is insane here. $0.50 per minute. That’s $30 an hour. An outsourced human call center worker costs about $0.017 a minute (roughly $1.02/hour). You’re paying a 30x premium to talk to an AI Agent. Everything seems good till it isn't.
166
Mar 31
Wait what..
Claude code source code has been leaked via a map file in their npm registry! Code: pub-aea8527898604c1bbb12468b…
2
123
Behfar retweeted
Mar 20

166
562
3,446
1,331,848
Mar 20

ALT Michael Jackson Comendo Picoca GIF

2
113