Accelerating entelechy and fighting stagnation. Design and Product Leadership @tiny_fish, “agentic engineer”
Joined February 2008
- Tweets 4,578
- Following 4,107
- Followers 748
- Likes 1,907
101 Photos and videos
Mario ⌘ retweeted
Jun 13
Introducing the Fusion API, the smartest compound model in the market.
Fusion achieves Fable-level intelligence at half the price.
How it works 👇
701
1,763
14,817
5,953,667
Mario ⌘ retweeted
Jun 12
it's been amazing to see the response to this 🖤 i'm reminded of one of my favorite quotes. in this age of AI, i think this is more relevant than ever.
having spent much of my career working in bigtech, it's a special opportunity to create something different — usually, we are optimizing for making things that 2 billion people find inoffensive, not creating beauty.
polished, perfect output is no longer interesting. humans, with all their flaws and quirks, are.
1
1
12
494
Mario ⌘ retweeted
Jun 1
🆕Grok Imagine’s Video Agent Moment: Cosmos, xAI, World Models, Generative UI, & the Codex Phase for Video!
latent.space/p/video-agents
@EthanHe_42, former @xai world model lead and @nvidia Cosmos researcher, explains why AI video may follow the same path as coding agents, how Grok Imagine went from zero to one, why text-to-video is only the autocomplete phase, how world models become real-time and interactive, why language models may become the control layer for video, and why the future of AI video may look less like a prompt box and more like an agent with a camera, editor, timeline, and tool belt.
5
9
45
48,447
Mario ⌘ retweeted
May 21
The most popular way to interpret AI is missing the bigger picture.
Models think in curved shapes. But sparse autoencoders (SAEs) work with straight lines.
Can they still capture models’ curved neural geometry? Yes, but not how you might think! (1/7)
May 7
Neural networks might speak English, but they think in shapes.
Understanding their rich *neural geometry* is key to understanding how they work – and to debugging and controlling them with precision.
Starting today, we’re releasing a series of posts on this research agenda. 🧵
25
151
1,017
173,724
Mario ⌘ retweeted
May 7
claude was a shape rotator all along
May 7
Neural networks might speak English, but they think in shapes.
Understanding their rich *neural geometry* is key to understanding how they work – and to debugging and controlling them with precision.
Starting today, we’re releasing a series of posts on this research agenda. 🧵
8
27
1,188
62,189
Mario ⌘ retweeted
May 12
New post 📝
A deep dive into a topic I wanted to explore for a long time: how to render a realistic sky and atmosphere
I explore everything from simulating the interaction of light with air, to handling sunsets, all in real-time, from ground to space
blog.maximeheckel.com/posts/…
ALT On Rendering the Sky, Sunsets, and Planets
35
102
857
59,680
Mario ⌘ retweeted
May 11
People talk, listen, watch, think, and collaborate at the same time, in real time. We've designed an AI that works with people the same way.
We share our approach, early results, and a quick look at our model in action.
thinkingmachines.ai/blog/int…
464
1,961
15,788
7,755,362
Mario ⌘ retweeted
May 8
Someone said our free Search and Fetch API is a marketing gimmick coz of the rate limits.
We took it personally.
So we 5x'd our rate limits.
Your agent doesn't sleep and neither do we.
Grab your API Key now: agent.tinyfish.ai/api-keys
Sound on 🔊
19
11
195
149,903
Mario ⌘ retweeted
May 7
Neural networks might speak English, but they think in shapes.
Understanding their rich *neural geometry* is key to understanding how they work – and to debugging and controlling them with precision.
Starting today, we’re releasing a series of posts on this research agenda. 🧵
307
1,674
11,184
3,162,178
Mario ⌘ retweeted
Apr 24
This is a glimpse of big changes ahead of us. If you’re betting on big central models you should think twice.
I run the exact same setup (M5 MacBook, qwen3.6-27B, pi, ollama) and while its not as fast or good as one of the big central models, it’s past the line of “cool demo” into “truly useful.” Kind of where the big frontier models were in late 2025.
In ~24 months we might have local models that are fast and good enough for most tasks.
Apr 24
This is where we are right now. And i’m not gonna lie it feels pretty magical 🧚♀️
Qwen3.6 27B running inside of Pi coding agent via Llama.cpp on the MacBook Pro
For non-trivial tasks on the @huggingface codebases, this feels very, very close to hitting the latest Opus in Claude Code, or whatever shiny monopolistic closed source API of the day is.
In full airplane mode.
Most people haven’t realized this yet.
If you have, it means you have a huge headstart to what I call the second revolution of AI.
Powerful local models for efficiency, security, privacy, sovereignty 🔥
40
54
902
109,425
Mario ⌘ retweeted
Apr 13
Workshop Labs is joining @thinkymachines.
We believe there's a path for AI to make humans matter more. We couldn’t be prouder to join Thinking Machines to see this work through.
workshoplabs.ai/blog/wsl-joi…
31
42
538
389,811
Mario ⌘ retweeted
Apr 12
A bunch of people have written me back saying this was the best newsletter I have ever sent (flattering) ... so here it is for those who don't subscribe: AI Is Not a Labor Crisis. It Is a Meaning Crisis.
68
155
1,404
275,828
new OS still in beta. already 100x better than our last version. not another openclaw — it's an OS designed for human-agent collaboration. safer, more elegant, simpler. go home, hand your machine to it, sleep well.
comment rt, giving away 50 invite codes.
62
30
161
129,805
Mario ⌘ retweeted
Feb 27
Mission Control from @FactoryAI
Looks like they’ve cracked the UX for autonomous running agents
I imagine the pour out from Cursor from Enterprises will be going to Factory
Excited to see what @AmpCode do in this space
There real value in being model agnostic
6
7
93
12,984
Mario ⌘ retweeted
Thank you @anthropic for the award and recognition. I hope this hackathon will enable other builders that had their big idea in the back of their heads. The technological barrier between domain expertise and creating a working solution has never been smaller.
@claudeai Code 1st Birthday & Hackathon showcase today in @SHACK15sf. Looking very much forward to meeting everyone!
🥉 postvisit.ai by Michal Nedoszytko
Patients leave doctor's offices every day without understanding their diagnosis.
Postvisit (built by a cardiologist) turns visit transcripts and medical records into ongoing, personalized health guidance.
62
109
1,673
255,836
Mario ⌘ retweeted
Feb 20
so your telling me a FULL-TIME DOCTOR vibe coded an entire post-visit care system to help patients understand their health and be compliant with meds.
and then your telling me he traveled across the globe to SF and won the Anthropic hackathon.
and now your telling me that this is the greatest time to build your ideas??
oh ok i get it now.
Feb 20
Cardiologist wins 3rd place at Anthropic's hackathon. Out of 13,000 applications. Built in 7 days by Michał Nedoszytko MD. Coded day and night - in the hospital, in the cloud, while flying from Brussels to San Francisco.
A few years ago, it would have been impossible for a doctor to build this alone in just a couple of days. AI changed that.
The project is called postvisit.ai. It is an AI agentic care platform for patients. Including reverse AI scribe it is a companion that guides the patient from the moment they leave the doctor's office.
Powered by the massive context window of Opus 4.6, it allows patients to explore their full medical history, connected devices, Evidence Based resources and external data sources — all in one place.
Today, the barrier to entry has vanished; even a practicing physician can build an application from scratch.
30
62
1,153
195,434
Mario ⌘ retweeted
Feb 18
If your Claude Code runs are longer than 40 minutes, then you're in the top 0.1% of users
Feb 18
Most Claude Code turns are short (median ~45 seconds). But the longest turns show where autonomy is heading.
In three months, the 99.9th percentile turn duration nearly doubled, from under 25 minutes to over 45 minutes. This growth is smooth across model releases.
34
15
387
46,140
Mario ⌘ retweeted
Feb 12
tinyfish web agent just scored 90% on mind2web bench
outperforming gemini by 21 points, openai by 29 and anthropic by 34
and we published every single run - all 300 tasks ran in parallel - in a public spreadsheet
check out our runs, and try them yourself 👇
80
64
481
386,018
