This is a fantastic explainer. Reach out to me or your Roboflow contact if you're a Roboflow user and want early access to some related features.

We shipped something pretty cool today at Roboflow but haven't made much noise about it. You'll see it when you login.

Big week for bringing AI to the physical world, I'd say. Very excited to be at a company in the middle of it.

This is soooo good. Realtime video moderation on Vision Agents using Roboflow Neural Architecture search, what a cool combination of bleeding edge tech.

"Alexa show the time" "I'm not sure how to help you with that" incredible

Some brief thoughts on Mythos We’ve known this was coming for a long time. At least, we *should* have. Extremely effective software vulnerability discovery was clearly coming to anybody paying attention. It has also been clear that all AI policy so far has been made and

This is a cool feature, and if you squint you can see where we're going with this for customers.

When Dr Satya Mallick is happy, we are happy Our paper on Neural Architecture Search (NAS) was accepted to ICLR 2026 and the way we trained RF-DETR to be SOTA We offer a hosted version NAS in-platform, it trains ~5k models for you and finds all the pareto optimal models for

btw this is available to all users, free and paid plans free plans come with $60 in credits so you can try NAS to see if you get a perf boost yes, free money and access to cloud GPUs glhf

Give the robot the knife untethered you cowards 😂

Great game. Got dragged to an arcade in SF to play this for the first time recently and had a blast.

At work we use our Structured Log of All Communication and Knowledge (Slack) the way it was intended: a giant human readable pubsub. Hundreds of messages a minute from automations. We default to open - public channels for 1:1 conversations are common. Years of context already.

Product Management is going to be full of philosophy and logic majors soon, and it's going to be exactly what's needed for products to continue to serve actual humans and not just be optimization problems for agents to recurse on. Right?

RF-DETR just got even better when training RF-DETR in roboflow, you can now use neural architecture search (NAS). esencially, we train ~6000 different models and pick the one that best fits your hardware and target latency. ↓ how it works?

Anyone else seeing LLMs absolutely unable to do times or dates today? Absolute rubbish across Claude and Gemini.

Unbundling services via MCPs and CLIs is going to kill a lot of loss leader strategies and move everyone to usage based billing sooooooo fast