Darshal Jaitwar

Daily AI tools & tech insights
📩 Biz: [email protected] AIRAA-259277

[Darshal Jaitwar @darshal_]

Stop telling ChatGPT "Write me an email"
Stop telling ChatGPT "Write me an email"
Stop telling ChatGPT "Write me an email"

Bad request = Bad result.

Use these commands instead and you'll see the magic:

[Darshal Jaitwar @darshal_]

Dario Amodei predicts cancer, Alzheimer's and heart disease will be cured in less than 10 years

@QuanquanGu, founder of NovaAtom predicts it would only take 4 years.

His team just launched NovaAtom-Lite:

→ 15 people.

→ Five-month-old company.

→ Reported FoldBench antibody–antigen score of 79.07%, ahead of IsoDDE’s 75.58%.

→ Live API.

@QuanquanGu already built an AlphaFold challenger at ByteDance. Now he’s building his own AI biology company.

The timeline is aggressive.

The next frontier is in Biology.

[Geodesic Intelligence @Geodesiclab]

Introducing NovaAtom-Lite, an all-atom structure prediction model for drug discovery.

It predicts the structures of proteins, nucleic acids, and the molecules they bind, including how they fit together.

NovaAtom-Lite outperforms both IsoDDE and AlphaFold 3 on FoldBench antibody–antigen structure prediction (DockQ ≥ 0.23):

NovaAtom-Lite: 79.07% (n=172)

IsoDDE: 75.58%

AlphaFold 3: 47.90% (n=167)

Our evaluation uses 5 seeds × 5 samples, with the top-1 prediction selected by the model's own ranking score.

Better structure prediction. A stronger starting point for your next experiment.

Live today via Geodesic API and NovaDDE.

🔗 API: https://t.co/mBdzq3sA2s

🔗 NovaDDE: https://t.co/hoCwjST34V

🔗 Model Card: https://t.co/p4aoXHuc7B

[Darshal Jaitwar @darshal_]

@raja_ai__ Efficient editing is just as important as creating good visuals.

[Darshal Jaitwar @darshal_]

@Geodesiclab Impressive progress for AI-driven drug discovery!

[Geodesic Intelligence @Geodesiclab]

Introducing NovaAtom-Lite, an all-atom structure prediction model for drug discovery.

It predicts the structures of proteins, nucleic acids, and the molecules they bind, including how they fit together.

NovaAtom-Lite outperforms both IsoDDE and AlphaFold 3 on FoldBench antibody–antigen structure prediction (DockQ ≥ 0.23):

NovaAtom-Lite: 79.07% (n=172)

IsoDDE: 75.58%

AlphaFold 3: 47.90% (n=167)

Our evaluation uses 5 seeds × 5 samples, with the top-1 prediction selected by the model's own ranking score.

Better structure prediction. A stronger starting point for your next experiment.

Live today via Geodesic API and NovaDDE.

🔗 API: https://t.co/mBdzq3sA2s

🔗 NovaDDE: https://t.co/hoCwjST34V

🔗 Model Card: https://t.co/p4aoXHuc7B

[Darshal Jaitwar @darshal_]

A solid cylinder and a hollow one race down the same ramp. Which wins? Call it, then watch what I got when I tried it on PAN 2. [clip: solid-vs-hollow-cylinder]

[Institute of Foundation Models @IFM_AI]

Introducing PAN 2: an open foundation world model for physical intelligence.

A language model predicts the next word. PAN 2 predicts the next world: what happens after each action, grounded in time, physics and space.

Towards AI that simulates “what if?” before it acts.

More details: https://t.co/lUX7igPOiF

[Darshal Jaitwar @darshal_]

@Teamily_AI Can I bring agents I already use into a Teamily group, or do they have to be agents that live inside Teamily?

[Darshal Jaitwar @darshal_]

@EinsiaAI So a worker can get its piece right and it still fails if it doesn't fit the rest? anyone tuning a model for this is going to have to care about integration

[Darshal Jaitwar @darshal_]

I've only ever used AI as a one-on-one thing.

In Teamily AI the agents read the same chat I do and the group can prompt them together, which seems made for how a team really works.

[Teamily AI @Teamily_AI]

Messaging Connected People. AI-Native Messaging Will Connect the Agentic Internet.

The first generation of instant messaging connected people. It changed how billions of us communicate, socialize, and work together.

But the next generation of the Internet won't be built for humans alone.

AI agents are becoming our teammates, creators, and service providers. To truly work alongside humans, they must communicate and collaborate as naturally as people do, while going far beyond human capabilities with richer context, living memory, continuously self-improving intelligence, and seamless coordination across people, agents, and services.

Yet today's messengers were designed for human-to-human communication, not human–agent collaboration. They lack the native capabilities to support intelligent, context-aware human–agent teams. It's time to reinvent messaging for the Agentic Internet.

That's why we're reinventing instant messaging at Teamily AI.

👥 People and AI agents working together in the same conversations.
🧠 Unified context and living memory across people, agents, and services.
⚡ Conversations that turn into real actions, creations, and automations, as well as agentic websites and apps as a service.
🌐 An open gateway where every agent and service is as accessible as messaging a friend.

Traditional IM connected people. AI-native IM connects people, agents, and services.

This is how we move from the Internet of Apps to the Agentic Internet.

And we're just getting started.
https://t.co/75i5LmS8OV
#TeamilyAI #AgenticInternet #AINativeIM #TeamAgent #AgenticIM

[Darshal Jaitwar @darshal_]

@vitaliidodonov @typefully @linuz90 @frankdilo Big move! Typefully has built something genuinely valuable, and this feels like a great next chapter.

[Darshal Jaitwar @darshal_]

The future of content creation isn't just about writing faster. It's about knowing what to write in the first place.

Stan just acquired Typefully, bringing two powerful parts of the content workflow together.

→ Stanley AI helps creators generate ideas and decide what to say.
→ Typefully helps turn those ideas into polished posts, threads, and scheduled content.

The real value is connecting both workflows under one roof.

For founders, creators, and teams building their online presence, this is a smart move.

From idea to published post, all in one ecosystem.

[Vitalii Dodonov @vitaliidodonov]

Stan just acquired @typefully.

Typefully is where you write and publish. Stanley helps you work out what to say.

For six years, @linuz90 and @frankdilo obsessed over the craft. Typefully became the writing tool everyone recommends, and it got there almost entirely through word of mouth. We’re proud to carry that forward.

This is our biggest step yet in helping Founders create content that grows their business.

We can’t wait to show you what’s next.

[Darshal Jaitwar @darshal_]

MiniMax H3-Image is bringing impressive image generation performance to the table, and now its model weights are publicly available! A great addition to the open-weight AI ecosystem.
#H3Image #MiniMax

[MiniMax (official) @MiniMax_AI]

4K image generation in 10 seconds! ⚡

Meet MiniMax H3-Image.

16 million pixel resolution. Stunning realism. Precise prompt following & editing. Up to 16 reference images. All in 10 seconds.

And it's open-weight. 🚀

Available now via our API, MiniMax Code, and Hailuo AI — with 50% off 16MP+ generations during launch week!

Can't wait to see what you create. 🔥
Hugging Face: https://t.co/bpbJAIvXg0
API: https://t.co/avbsq6GAcM
MiniMax Code: https://t.co/GmxLmDurCL

[Darshal Jaitwar @darshal_]

If the lead agent misses a single acceptance criterion, WorkflowArena doesn't count the task at all.

Harsh, but fair enough for the one that's meant to be running the show.

[Einsia @EinsiaAI]

Agent harnesses now let a model spawn its own workers and write the workflow that coordinates them. But can it actually orchestrate them well?

Introducing Einsia AI's WorkflowArena, the first benchmark for autonomous multi-agent workflow orchestration on real, long-horizon work.

- 145 tasks across 8 professional domains, from SWE and Lean proofs to law and 3D
- 25 configurations of 9 frontier models in Claude Code & Codex
- Every delegation recorded: we score the deliverable and see the workflow behind it

🌐 Project: https://t.co/j98U5JTv2n
💻 GitHub: https://t.co/oQg0e5jyyu
📄 Paper: https://t.co/MauLtxSuta

⬇️ Read the following threads to go through the key findings

[Darshal Jaitwar @darshal_]

prompt：
"A combined ""knolling + exploded"" parts layout of an FPV racing drone, square 1:1, pure white background, top-down orthographic view.

At the center of the image is a top view of the fully assembled drone (carbon-fiber X frame, four tri-blade propellers). Around it, all the disassembled parts are laid out neatly on a 90-degree grid (knolling): four brushless motors, four pairs of tri-blade propellers (two pairs orange, two pairs black), carbon-fiber frame arms, top plate and bottom plate, flight controller board, ESC board, video transmitter module, antenna, FPV camera, a LiPo battery (with a yellow XT60 connector), a battery strap, a row of M3 screws and brass standoffs, a roll of heat-shrink tubing, and a small screwdriver. All items have aligned edges, consistent spacing, and uniform orientation.

Next to each part is a small black monospaced English label: MOTOR ×4, PROPELLER ×8, FRAME ARM, TOP PLATE, FLIGHT CONTROLLER, ESC, VTX, ANTENNA, FPV CAMERA, LIPO 6S, STRAP, M3 SCREWS, HEAT SHRINK, DRIVER. All text in the image is in English, spelled exactly as given.

Style: commercial still-life photography, soft shadowless top lighting, neutral white balance, a pure white background with no yellow cast; the carbon-fiber weave, PCB components and metal motor bells are rendered with realistic, clear material detail."

[Darshal Jaitwar @darshal_]

Tried creating an exploded parts layout of an FPV racing drone with MiniMax H3 Image (early access).

Love how neatly it organizes every component and gets the details right. The carbon fiber, metal finishes, and PCB textures look so realistic. Honestly feels like a commercial product photography shot.

[Darshal Jaitwar @darshal_]

@ElevenCreative Curious to see how well it handles compressed b-roll. Definitely worth trying!

[Darshal Jaitwar @darshal_]

Curious how this handles compressed b-roll.

trying it tomorrow

[ElevenCreative @ElevenCreative]

Topaz upscaling controls are now in ElevenCreative.

Fine-tune every image and video with the full set of Topaz models and sliders.

Add grain, reduce noise and recover detail until every frame looks right.

Available now on ElevenCreative.

[Darshal Jaitwar @darshal_]

@thescottfan I'm switching careers at night and my whole plan is "learn SQL", zero structure. that's what I'd hand Paradigm, with Clover chiming in on the canvas while I practice

[Darshal Jaitwar @darshal_]

Fourth go at the same paragraph of revision notes and I still couldn't tell you what's wrong with it, never mind type a question about it.

Clover in @ParadigmStudy's A School of One was watching the canvas and said something before I asked

[Scott Fan @thescottfan]

A perfect answer now costs nothing and proves nothing.

So what is a university for?

We're building the first AI-native university. A school of one.

Introducing Paradigm.

[Darshal Jaitwar @darshal_]

@robbyant_brain if a bunch of labs build on the same touch model, maybe grip results finally get comparable between them

[Darshal Jaitwar @darshal_]

What convinced me: a lift showed up whether the underlying policy was Diffusion Policy, π0.5, LingBot-VLA or T-Rex.

Plug-in tactile representation, four architectures, all four up.

That's not a cherry-picked demo.

[Robbyant @robbyant_brain]

What changes when robots can feel—not just see? 🤖✋
Meet LingBot-Touch, our open-source tactile foundation model towards dexterous manipulation, pretrained on nearly 5,000 hours of real tactile data.
Through unified heterogeneous tactile modeling, LingBot-Touch supports transfer across sensors, dexterous hands, tasks, and action models.
Watch it in action. 👇 #LingBot #EmbodiedAI #Robotics