
I help you make money with AI l AI Tech Tools | Robotics Engineering | Helping you to leverage AI to grow | Sharing Insights of AI l ✉️ [email protected]
United States
Joined February 2026
- Tweets799
- Following132
- Followers50.4K
- Likes1.4K
Pinned Tweet
What if your AI coding agent could actually own the task?
Not just write code when you ask.
Qoder can take a goal, break it down, research what’s needed, use tools, edit files, run tasks, and verify the result.
It feels less like chatting with an AI and more like working alongside one.
Bonus: Qwen3.8-Flash is currently FREE at 0.0x Credits.
Let’s break it down 🧵👇
🤝 Paid partnership
Medical AI is moving beyond the prompt-and-response experience.
LingSante-3.1-air-VL can gather visual information, reason through findings, search for supporting evidence, and interact with apps.
That's a much more interesting vision for AI-powered healthcare workflows.
Meet LingSante-3.1-air-VL, an open-source multimodal medical model with 124B total and 5.5B active parameters. It combines medical image understanding, report generation and medical reasoning with multimodal search and tool use.
In our evaluation, it leads the compared models on AFUMED-Derm, AFUMED-Report, MIMIC-CXR and medical VQA (averaged across benchmarks), ahead of GPT-6.1 Sol, Gemini 3.8 Flash on all four.
It also demonstrates strong general-purpose capabilities, scoring higher than GPT-6.1 Sol on BrowseComp-VL and Gemini 3.8 Flash on AndroidWorld, alongside competitive performance on MMSearch. Explore LingSante-3.1-air-VL and build with it.
1
1
4,659
Healthcare AI is getting another serious contender.
LingSante-3.1-flash delivers strong benchmark results and combines them with complex medical reasoning, research capabilities, and long-horizon workflows.
Definitely worth exploring.
Today, we release LingSante-3.1-flash, an open-weight MoE model with 560B total parameters and 25B activations per token. It unifies diverse medical capabilities while retaining general capabilities, including complex medical reasoning, long-horizon workflow execution, and professional behavior alignment.
LingSante-3.1-flash moves beyond medical QA toward real-world healthcare problem. The model leads on HealthBench Professional (65.72), HLE (text-only, with tools: 72.22) and BrowseComp (with context: 94.10), outperforming GPT-6 Astra, Claude Opus 5.5 and Gemini 4 Argon on all three. LingSante-3.1-flash also performs well on professional long-horizon medical workflows, such as PhysicianBench (57) and EHR-Complex (74.40).
Together, these results establish LingSante-3.1-flash as the leading open-weight models for healthcare.
The weights are available. Explore LingSante and build with it.
More below ↓
1
1
4,698
Robin retweeted
Replying to @MyClaw_Official
Once people expect ownership, they stop accepting tools that only answer.
Robin retweeted
The assistant-to-teammate shift is a change in expectations, not just vocabulary.
X News just put MyClaw next to Google, Salesforce, Vercel, and Instinct.
We didn't ask for the seating chart.
We shipped ready-to-work AI agent teams — not another blank chat you have to babysit.
2.2K posts in a few hours.
The category isn't prettier assistants. It's teammates that already have tools, context, and a computer.
That's what we built. That's what people are talking about.
Robin retweeted
Not every creative session needs a serious plan.
Sometimes I just want to open CapCut Web, play around with a few prompts, and see what kind of visuals I can come up with.
➪ capcut.com/my-edit?utm_mediu…
#CapCutWeb #CapCutWebVideoEditor
🤝 Paid partnership · 🤖 Made with AI
Not every creative session needs a serious plan.
Sometimes I just want to open CapCut Web, play around with a few prompts, and see what kind of visuals I can come up with.
➪ capcut.com/my-edit?utm_mediu…
#CapCutWeb #CapCutWebVideoEditor
🤝 Paid partnership · 🤖 Made with AI
One idea can turn into five different styles. A simple concept can become something totally unexpected.
That's the best part of experimenting: you never know which result will spark your next project.
🤝 Paid partnership · 🤖 Made with AI
1
51
And there's more room to play from Nov 20–26.
New or upgrading CapCut Web Ultra users can enjoy 7 days of unlimited image generation on select models.
Your next creative rabbit hole starts here. 👀
#AIImage #CapCutWebDesignStudio
🤝 Paid partnership
1
4,239
Robin retweeted
The worst follow-up gaps start with an out-of-office reply nobody reads properly
Reevo went through 170,000+ sales emails and calls, and the advice is simple: pause until they're back, then check if their backup contact is even the right person.
🤝 Paid partnership
The best sales teams win 5x as often.
We analyzed $1.2B+ in pipeline, 10,000+ deals, and 170,000+ emails and calls on @reevo_ai to find out how they do it.
Check it out below 👇
Hy4 flash is now on MyClaw, and the launch thread has a few prompts worth borrowing.
🤝 Paid partnership
Hy4 flash is live on MyClaw ⚡
>Sketch a character
>Turn a floor plan into a Blender scene
>Research a topic
Try Hy4 flash → myclaw.ai
Robin retweeted
The vibe: cruising through a fictional city with nowhere to be and nowhere to rush. Just a car, a sunset, and an endless city skyline.
I described that feeling to ChatGPT and asked it to use the CapCut plugin to generate a GTA-style video with Seedance 2.5. Pretty cool to see the idea take shape.
See what GPT-6.2 Astra can do through the CapCut plugin. Your first month of CapCut Pro is 90% off this Black Friday!
capcut.com/tools/capcut-x-co…
#GTA6 #ChatGPT #CapCutDidThat #CapCut62Astra #CapCutChatGPTPlugin
🤝 Paid partnership · 🤖 Made with AI
The assistant-to-teammate shift is a change in expectations, not just vocabulary.
X News just put MyClaw next to Google, Salesforce, Vercel, and Instinct.
We didn't ask for the seating chart.
We shipped ready-to-work AI agent teams — not another blank chat you have to babysit.
2.2K posts in a few hours.
The category isn't prettier assistants. It's teammates that already have tools, context, and a computer.
That's what we built. That's what people are talking about.
Robin retweeted
If I were giving this to students, I'd have them document which parts of the scene the photos actually support.
🤝 Paid partnership
Gemini 4 Neon is live on MyClaw ✦
Ask your agent to turn a few photos into a 3D scene you can explore in your browser.
Neon can read the photos and write the code, with up to 1M tokens of context.
Explore Neon → myclaw.ai/
Robin retweeted
The real bottleneck in AI systems might be general-purpose thinking.
ForgeSystem takes a different approach: build for one workload, remove unnecessary boundaries, and make operations work together.
Three kernel passes become one. Entire decoding steps become megakernels.
Sometimes, the biggest performance gains come from doing less, but doing it smarter.
🤝 Paid partnership
Why does ForgeSystem forge products faster just by building it for a single workload?
In a general-purpose framework, most boundaries between modules exist for compatibility rather than speed. When ForgeSystem builds for one specific demand, it drops those boundaries and organises separated operations around the same hardware resources.
In a standard training loop, gradient clipping, the Adam update, and precision conversion are three separate passes over the parameters — three kernel launches, three round-trips to memory. ForgeSystem merges them into a single kernel sharing one data path. For inference, it pulls an entire decoding step including compute, data loading, on-chip storage, and sync into one megakernel so intermediate write-backs disappear.
Alongside deep specialisation and large-scale exploration, this vertical integration is where ForgeSystem gets its gains: 1.047–1.332× MFU on training, 1.21× over SGLang on decoding, and matching or beating expert libraries in 18/22 kernel tasks.
More about deep specialisation & large-scale exploration: arxiv.org/html/2611.12288v1
Robin retweeted
.AI agents are no longer just something you talk to. They're becoming something that can actually see what you see.
Giving every Teamily agent a face, real-time vision, and camera access makes the whole experience feel like a step closer to having an AI companion in the real world. Pretty exciting!
#TeamilyAI #AIGlass #PhysicalAgent #VideoAgent
🤝 Paid partnership
Introducing Teamily Video Agent
We give all Teamily agents a face.
They also see the world you see through your AI glasses and your camera.
#VideoAgent #AIGlass #PhysicalAgent
Robin retweeted
GTA 6 just came out and I'm already using GPT-6.2 Astra in ChatGPT alongside CapCut to build first-person sports car runs.
Good thing CapCut is running 90% off your first month for Black Friday right now here is the exact flow.
➪ capcut.com/tools/capcut-x-co…
#GTA6 #CapCutDidThat
🤝 Paid partnership · 🤖 Made with AI
The worst follow-up gaps start with an out-of-office reply nobody reads properly
Reevo went through 170,000+ sales emails and calls, and the advice is simple: pause until they're back, then check if their backup contact is even the right person.
🤝 Paid partnership
The best sales teams win 5x as often.
We analyzed $1.2B+ in pipeline, 10,000+ deals, and 170,000+ emails and calls on @reevo_ai to find out how they do it.
Check it out below 👇
The vibe: cruising through a fictional city with nowhere to be and nowhere to rush. Just a car, a sunset, and an endless city skyline.
I described that feeling to ChatGPT and asked it to use the CapCut plugin to generate a GTA-style video with Seedance 2.5. Pretty cool to see the idea take shape.
See what GPT-6.2 Astra can do through the CapCut plugin. Your first month of CapCut Pro is 90% off this Black Friday!
capcut.com/tools/capcut-x-co…
#GTA6 #ChatGPT #CapCutDidThat #CapCut62Astra #CapCutChatGPTPlugin
🤝 Paid partnership · 🤖 Made with AI
Robin retweeted
Imagine an AI confidently inventing clinical data and sending researchers into weeks of unnecessary experiments.
That's the real danger of hallucinations in science.
Santé's ability to recognize uncertainty is what makes this interesting. In drug discovery, honesty about missing evidence can be just as valuable as intelligence.
When your AI says “I don’t know,” it costs you a few minutes with the paper. A plausible wrong answer could cost you weeks in the lab.
While testing the @AntLingAGI Santé health model on our drug rescue engine, an encouraging sign was its ability to admit uncertainty 🧵












