A collection of molecules temporarily hanging out together // Head of RLE @ Handshake AI
Brooklyn, NY
Joined December 2017
- Tweets797
- Following335
- Followers251
- Likes2.2K
Brendan M. O'Connell retweeted
Is Opus 5.5 a good Vision-Language Model? We ran it on ATLAS Visual Life Sciences (VIALS), 161 visual interpretation tasks from professional life sciences workflows.
It refused 15% of requests, but it (and GPT-6 Astra) still mark a major jump in VLMs for real scientific work.
Brendan M. O'Connell retweeted
SEPT ’26: HUMANS STILL VERY MUCH HAVE JOBS.
AI isn't even close to meeting the standard required of the most junior professionals in banking, advisory, and private equity today.
ATLAS Finance agents must navigate complexity like a human does: "Johnny you just got staffed - check your email."
Human: 100%.
Best AI: 12%.
Agents have access to data rooms, email, chat, calendars, docs, notes, Excel, etc. Our real finance professionals pushed the realism frontier further by building dozens of coworker and client personas that introduce ambiguity, (controlled) contradictions, and real-time updates that must be adjudicated to successfully complete the client-ready deliverable.
We gave 11 frontier models 100 expert-level tasks that each take humans 15–30 hours.
On Wall Street there's a saying, "If it's 95% right, it's 100% wrong." Autonomous knowledge work outside software still has a very long way to go.
Brendan M. O'Connell retweeted
Fleet’s reported run rate went from ~$1M to $63M in six months. Mercor’s went from ~$500M to $2B in under a year. There aren’t many AI markets growing this fast.
But the product also expires unusually fast. Labs moved from RLHF to expert SFT to RL environments, and are now looking for real company workflows. A vendor can be growing 50x and still be late to the next category.
@patrick_tianyi and I spent the past few months talking to founders, researchers, and buyers about where the money is going, including a part of the market that has barely been written about in English: China’s new research-led data companies.
Here’s what we learned ↓
Brendan M. O'Connell retweeted
AI data factories like Handshake, Scale, Surge, and Mercor have a big role to play in the advancement of AI at the frontier and in the enterprise. This is my full thesis.
Brendan M. O'Connell retweeted
Introducing Patch Radius.
Spark a chain reaction of climate impact throughout your network of clients, portfolio companies, or supply chain. patch.io/radius
Fun fact - Luka's 60 point performance is the 60th highest points scored in a single game by an NBA player (per: basketball-reference.com/lea…)
WeWork stock price completely misrepresents the value of their indoor plant collection. I would propose we take them private, sell the plants, then re-IPO. Who's in?
"The next time you experience a sliver a joy, grab on to it, celebrate it, appreciate that moment. You may even find that there’s more than a sliver hiding beneath the surface"
~~ some wisdom that flowed from my mind to a friend today ~~
Was at the gym in Paris this past weekend. Asked the manager if they had any protein. She said ouais
Brendan M. O'Connell retweeted
Our ETHmissions panel yesterday was fantastic and informative. Many thanks to our brilliant guests @UliGall from CCRI and @oconnell__ from @usepatch!!
...Unfortunately, the link I shared didn't work for many 😭 apologies
Luckily, it was recorded: here's the link to it
I think we can all agree that Serbia needs to work on their creativity when it comes to naming babies
Nikola Jovic, NBA Draft prospect, is getting some shots up at the Nuggets pre-draft workout today.
(via @katywinge)
If you are starting a climate tech company and think it's a good idea to put the word "carbon" in your name, it's not
There's a town in Wales called Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch, which translates to “Saint Mary’s Church in a hollow of white hazel near the swirling whirlpool of the church of Saint Tysilio with a red cave”
