Adish Jain ☕️

5K posts

Adish Jain ☕️ banner
Adish Jain ☕️

Adish Jain ☕️

@_adishj

building video agents @mosaic_so & @motion_so / prev @ycombinator @tesla @awscloud @ucberkeley

SF Katılım Mayıs 2020
1.9K Takip Edilen4.6K Takipçiler
Sabitlenmiş Tweet
Adish Jain ☕️
Adish Jain ☕️@_adishj·
introducing Motion, a video agent for tasteful motion design. this launch video was made entirely with Motion. 👇🏽 comment "MOTION" to get early access + free credits. tag @motion_so in any post on your X feed for a surprise. here’s how it works + examples (thread):
English
814
87
1.4K
204.5K
OpenAI
OpenAI@OpenAI·
GPT-5.5 Instant is more dependable, with significant improvements in factuality, especially in domains where accuracy matters most, like medicine, law, and finance. It’s also stronger across everyday tasks, from analyzing image uploads to answering STEM questions to knowing when web search can help.
OpenAI tweet mediaOpenAI tweet media
English
9
15
367
20.4K
OpenAI
OpenAI@OpenAI·
GPT-5.5 Instant is starting to roll out in ChatGPT. It’s a big upgrade, giving you smarter, clearer, and more personalized answers in a warmer, more natural tone. And it's also more concise, which we heard you wanted. We think you'll love chatting with it.
English
176
212
2.3K
95.1K
Brycent
Brycent@brycent·
The first LLM with a 12M token context window 🤯 Most people don't realize how monumental this is... SubQ just improved the size context windows. Most agents get cooked and start screwing up as the context window gets larger. But the entire economic model could flip: Because building agents that are smart, fast and affordable just became possible.
Alexander Whedon@alex_whedon

Introducing SubQ - a major breakthrough in LLM intelligence. It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA), And the first frontier model with a 12 million token context window which is: - 52x faster than FlashAttention at 1MM tokens - Less than 5% the cost of Opus Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention). Only a small fraction actually matter. @subquadratic finds and focuses only on the ones that do. That's nearly 1,000x less compute and a new way for LLMs to scale.

English
3
1
6
438
Adish Jain ☕️
Adish Jain ☕️@_adishj·
what do Ornadyne (YC P26)'s autonomous flying drones and Klaimee (YC P26)'s insurance for AI Agents have in common? they're some of the coolest companies in the batch, they both launched today, and @mosaic_so produced both their launch videos! since launching our new video agent, @motion_so, just two weeks ago, we've already helped six @ycombinator startups with their launch videos. and because we use our own video agent to help produce these, we can move extremely fast and for a fraction of the price of a traditional agency. we can do physical shoots in SF or completely motion-driven videos as well. if you're a founder in the current YC batch or are a founder looking for a launch video, DM me or leave a comment — we got you 🤝 congrats on the launches @geourgkivijian, @Armakelyan of Ornadyne (YC P26) and @inesboutem, @JulsCaton of Klaimee (YC P26)!!
Adish Jain ☕️ tweet mediaAdish Jain ☕️ tweet media
English
0
3
8
138
Adish Jain ☕️ retweetledi
Mosaic
Mosaic@mosaic_so·
Our first key feature released in May is Conditional Routing. But what can it actually do? It lets your workflow decide what happens next. Route clips based on quality, format, or signals automatically. One input. Multiple paths.
Mosaic tweet media
Mosaic@mosaic_so

Here are the 9 key features of the May product update for Canvas: 1. Conditional Routing: conditionally execute different workflow branches based on certain visual checks 2. Audio Syncing: upload multiple audio and video sources and auto-sync them. Perfect for podcasts, multi-cam shoots, and productions with multiple audio sources. 3. Updated Reframe: a completely new in-house implementation of Active Speaker Detection and reframing, fully editable in the timeline. 4. Image Generation: storyboard out entire scenes in the Canvas with Image Generation, powered by ChatGPT Image 2.0 5. Video Generation: go from Image Gen into Video Gen for image-to-video. Also supports video-to-video workflows to modify and alter existing videos. 6. Audio Generation: create new music, sfx, and sound design with a dedicated Audio Gen tile. 7. AI Avatar: now powered by Seedance 2.0 for highly realistic AI avatar and AI UGC workflows. 8. SlowMo tile: for cinematic scenes where the pace needs to be slowed. 9. Dropbox Triggers: just as with Google Drive and YouTube, now automatically ingest assets from designated Dropbox folders and export back after editing. As Canvas becomes more powerful, it becomes the perfect framework to start automating your video editing workflows. Try our visual workflow builder, API, and Agent Skills at mosaic [dot] so.

English
0
2
3
246
Adish Jain ☕️
Adish Jain ☕️@_adishj·
klaimee is one of the most interesting companies in the current @ycombinator batch bc insurance in the agent era is such an interesting concept what happens if your agent does something wrong? who owns the liability? Klaimee is building the trust layer that is necessary for agents to actually scale nice launch video and congrats on the launch team :)
Y Combinator@ycombinator

Klaimee (@klaimee_ai) is the insurance for your AI agents. They help your company deploy autonomous AI at scale, bridging the gap that cyber and E&O policies are excluding today. Congrats on the launch, @inesboutem & @JulsCaton! ycombinator.com/launches/QCC-k…

English
0
0
6
347
Theo Wayt
Theo Wayt@theo_wayt·
Scoop 🚀🖥️: Starcloud, a two-year-old orbital data center startup, is in talks to 2x its valuation to $2.2b just one month after announcing another round at $1.1b. It has launched a single satellite with a single Nvidia chip, but the SpaceX IPO is getting investors excited
Theo Wayt tweet media
English
14
16
177
54.9K
Jagadeesh
Jagadeesh@JagBuilds·
🛑🛑 Looking for a motion graphics designer for promotional video of my product Please suggest! 🙏🙏
English
1
1
1
44
Bridger Pennington
Bridger Pennington@bridger_penn·
Looking for good video promo video people, that can help create a @claudeai style video for launching our @fundlaunch_ai product Let me know if you know anyone good
English
3
0
1
165
Naval
Naval@naval·
New podcast is up: - The fully interconnected startup - Drones and the logic of violence - The coming hardware renaissance - Creators vs. Doomers Follow @navalpodcast
Naval Podcast@navalpodcast

New podcast is up! ‘Nothing Ever Happens’ Is Over The Fully Interconnected Startup 00:00 You Don’t Need the Explicit Intranet Anymore 04:14 May You Live in Interesting Times 06:55 Drones Democratize Violence 10:40 Biothreats Could Also Get Democratized 12:43 AI Interfaces Unlock Hardware 15:09 Optimism Requires Creativity 17:35

English
87
96
1.3K
197.6K
sheep
sheep@thefarmersheep·
what does the “pdf” in “pdf” file mean
English
61
3
113
9.1K
Irushi
Irushi@Im_IrushiK·
Guess the legend.
Irushi tweet media
English
508
646
17.6K
4.9M
Avi Patel
Avi Patel@avipat_·
Kled has been top 100 on the Nigerian app store for 4 months, we clearly don't need the promotion there. We also have no need to promote our app in a region that no longer has access to it. Kled is the number 1 app in many asian countries that have a flourishing user base. I have nothing against the people in Nigeria, we have a large community there and I felt obligated to share why we removed the app from this region to the small % of users who were genuine contributors. If you want to verify these stats you are welcome to here, screenshot attached as well. Pipe down. app.sensortower.com/app-analysis/c…
Avi Patel tweet media
English
10
3
489
58.5K
Avi Patel
Avi Patel@avipat_·
We have removed Kled from the Nigerian app store and IP banned the entire region. The first thing I would like to say is I have nothing against Nigeria. I have a ton of friends from this region and these were some of our earliest app adopters. Genuinely, thank you all for the support. Kled has been up and running and out of beta for 4 months now. We have paid out hundreds of thousands of people for their data, and our users have uploaded over 1 billion assets onto our platform. After several months of uploads we found that Nigeria had a ≈95% fraud rate. Instead of real, usable data, users were uploading pictures of black screens, duplicate photos, internet generated images, AI generated images, etc. at an unimaginable scale. In comparison, Malaysia, Indonesia, and the Philippines have a less than 10% fraud rate across 10x the userbase size. Our fraud system is fast to catch these issues but the level of complexity of these schemes is getting out of hand. This weekend we were flooded with thousands of fake Japanese passports and identity cards with Nigerians photoshopped onto them in our KYC system. That was the final straw. As a startup we can't afford to eat the costs of that data overhead, so we temporarily removed the app from the region while we improved our fraud detection and banning system to quickly filter out bad actors when the time is right. On top of all of this, every time we make a post there is someone asking us to bring the region back within seconds. We hear you, but it's gotten out of hand. We've made this decision with great care. We love everyone who has genuinely supported Kled from Nigeria, and we hope to return when the time is right. -Kled Team
Avi Patel tweet media
English
2.8K
1.7K
8.5K
5.2M
Adish Jain ☕️
Adish Jain ☕️@_adishj·
since launching our new video agent, @motion_so, last month, it feels like it's all we've been talking about. but while Motion is still in early access, @mosaic_so Canvas and API have quietly been hitting all-time usage highs. we’ve constantly been shipping major updates and new primitives into Canvas just to keep up with demand. here are the 9 key features of the May product update for Canvas: 1. Conditional Routing: conditionally execute different workflow branches based on certain visual checks 2. Audio Syncing: upload multiple audio and video sources and auto-sync them. Perfect for podcasts, multi-cam shoots, and productions with multiple audio sources. 3. Updated Reframe: a completely new in-house implementation of Active Speaker Detection and reframing, fully editable in the timeline. 4. Image Generation: storyboard out entire scenes in the Canvas with Image Generation, powered by ChatGPT Image 2.0 5. Video Generation: go from Image Gen into Video Gen for image-to-video. Also supports video-to-video workflows to modify and alter existing videos. 6. Audio Generation: create new music, sfx, and sound design with a dedicated Audio Gen tile. 7. AI Avatar: now powered by Seedance 2.0 for highly realistic AI avatar and AI UGC workflows. 8. SlowMo tile: for cinematic scenes where the pace needs to be slowed. 9. Dropbox Triggers: just as with Google Drive and YouTube, now automatically ingest assets from designated Dropbox folders and export back after editing. as Canvas becomes more powerful, it becomes the perfect framework to start automating your video editing workflows. try our visual workflow builder, API, and Agent Skills at mosaic [dot] so. --- this video was made entirely with our new video agent, @motion_so. For early access, sign up at motion [dot] so.
English
5
5
25
1.7K

Keşfet

@OpenAI @motion_so @brycent @mosaic_so @ycombinator @geourgkivijian @Armakelyan @inesboutem