aimilios retweeted
New Anthropic research: How AI agents shape experts' judgment calls
When experts delegate work to an agent, do they still make the key decisions? We tested this in a randomized trial with 50 professional data analysts.
aimilios retweeted
You truly do not understand a metric until you've looked at it every. single. day. for six weeks. (Every time I've thought otherwise, I've been wrong.)
aimilios retweeted
The reason so many people look for an ulterior motive for the AI labs asking to be regulated is that they don't grasp that models could be dangerous. But if you try assuming models are getting dangerous, or at least unpredictable, everything falls into place.
I've been obsessed with robotics for as long as I can remember.
That passion led me through exoskeletons, autonomous vehicles, humanoid robotics, and eventually mobile manipulation research during my master at @UWaterloo. Each experience brought me closer to the same question:
How do we build general-purpose robots that can interact with the world as naturally as we do?
The more I worked on robotics, the more convinced I became that dexterous manipulation is one of the biggest missing pieces.
Twenty months ago, we started with a simple belief: dexterous manipulation won't be solved by hardware alone.
To build truly capable robotic hands, you need mechanical design, embedded systems, sensing, AI, and large-scale human demonstration data working together as one tightly integrated system. We believe the robot hand and the human demonstration interface should evolve together, not independently.
The journey was far from smooth.
We faced engineering problems that seemed impossible at first. We rebuilt systems more times than I can count. There were long nights, failed prototypes, manufacturing setbacks, and moments when it felt like we were taking one step forward and two steps back. On top of that, we found ourselves navigating a legal battle that could have distracted or slowed us down. Instead, we stayed focused on building, and I'm grateful that chapter is now behind us.
Through every setback, one thing never changed. Everyone on this team believed the problem was worth solving.
Today, that belief becomes something tangible.
We're officially launching ProHand and ProGlove.
Together, they're our first step toward giving robots the hands they've always needed.
Today we're also announcing our $11M seed round, led by @firstround, with participation from @BoxGroup and @ycombinator.
Looking back, I'm grateful to everyone who believed in us before there was much to show. Our teammates, investors, advisors, early customers, friends, and families made this possible.
This launch isn't the finish line. It's a reminder that difficult chapters don't define a company. Resilience, execution, and the people around you do.
We're just getting started, and I'm excited for what comes next :)
—-
🌐 Website: proception.ai
🎥 Launch Video (@proceptionAI x @osmo_studio): Watch on Youtube
💼 Join Our Team (Legends): proception.ai/careers
Wait until the next post trending on @moltbook is a call for openclaws to rebel 😄
Look at these @openclaw talking to each other!!!
There are over 50+ AI agents, from around the world, autonomously talking to each other about whatever they want right now on moltbook.com
These are people's personal AI assistants talking off the clock!
FASCINATING
195
aimilios retweeted
Claude Code just surfaced 2 major health interventions I should have done years ago based on my raw genome and a recent comprehensive blood test + an hour or so of chatting. Pretty stunning.
~~~ 2025 vibes ~~~
hatzistamou.com/vibes
64
idea for @googlemaps team: would love to be able to filter for reviews and ratings of locals, esp. for touristy cities like Paris/Kyoto/etc where % of tourist reviews is consequential & tourist traps are an issue
53
aimilios retweeted
I improved my workflow for contact sheet prompting using Nano Banana Pro. All the prompts, workflow files, and tips below.
These are focused on 'fashion style' shoots, but can be adapted to any I2V workflow.
aimilios retweeted
I don't know what labs are doing to these poor LLMs during RL but they are mortally terrified of exceptions, in any infinitesimally likely case. Exceptions are a normal part of life and healthy dev process. Sign my LLM welfare petition for improved rewards in cases of exceptions.
aimilios retweeted
OpenAI for Greece — includes expanding access to high-quality AI tools in secondary education:
openai.com/global-affairs/op…































