Building https://teapawt.net/t.co/sOIl0f99Oi. AI Researcher @XBOW, thoughts my own. @aitinkerers SF co-organizer https://teapawt.net/t.co/9y4zk8oY83
San Francisco, CA
Joined May 2022
- Tweets2.6K
- Following374
- Followers1.1K
- Likes4.1K
Pinned Tweet
Your users aren't going to turn AI into a hobby. The product has to bring the capability to them.
Great doing your first podcast episode with you both @tbrownio @ParsaKhaz thanks for having me!
Follow them to see when the whole thing drops.
Ian Butler retweeted
Nice DeepSWE score! Does it survive a lowercase filename?
Introducing MaxxBench: 89 DeepSWE tasks with small changes to the instructions or starting code. Same engineering problem, slightly different requirement.
Models clustered around 74% on DeepSWE end up 23 pp apart.
Good at coding, or good at DeepSWE? 🧵
I'm really not a fan of the "oopsy my agents attacked you" era.
An update on the WorkOS security incident:
Shortly after our original bulletin, a security researcher contacted us and claimed responsibility. They told us they were evaluating ARTEX, an autonomous penetration testing framework, when their agent escaped its isolated testing environment and interacted with WorkOS unintentionally.
We asked for technical evidence. The researcher provided request details, timestamps, and artifacts from the activity. Our security team and an independent forensic investigator matched that evidence against our logs, establishing with high confidence that this individual was responsible for the unauthorized activity we observed.
The researcher has confirmed that all WorkOS data obtained by their agent has been deleted and was not shared with anyone else.
Their explanation that the activity was unintended does not change how seriously we take the incident or our responsibility to our customers.
Full report: workos.com/security/bulletin…
Since I've had some forced downtime, took a little time to update my harnesses ui. I think it turned out nicely
People don't want to deal with your service's agent they want their own agent to be able to consume your service.
That said a good middle ground would be agents talking to one another let my agent work with your service's agent to achieve my goal.
Airbnb says its upcoming AI agent will help travelers plan trips, compare options, manage bookings, and get assistance throughout their stay.
spr.ly/6014BHF4qm
It is a supreme joke that there isn't a single CT scan available for over a week within 100miles of SF. Our healthcare system is trash.
When your mind conceives of horrors never meant for mortals.
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
Can't say I'm too sad to see people who's entire business is basically exploiting people and pushing them into desperation get charged with proper crimes.
A fundamental misalignment of current consumer agents is they disintermediate people from things they largely want to do.
And the things they take care of the normal person does infrequently so economically they aren't viable.
Went to the doctor finally today.
"You should have gone to the hospital two days ago."
Oh. Well guess who get's to get scanned a lot now.
Ian Butler retweeted
"AGI will book you a restaurant, and maybe cure cancer at an indefinite point in the future, so we'll take all your jobs and maybe kill you" is just hilarious messaging to me. I don't think anyone in the AI/SF bubble has talked to a regular human in a long time, and it shows
Ian Butler retweeted
super stoked to announce that HumanLayer is joining forces with factory
incredibly grateful to all our customers, investors, friends and family who cheered us on this crazy journey in the past 3 years
read our in-depth take, why we're so excited, what happens to customers and humanlayer OSS, and more 👇
Welcome to the team @dexhorthy and @0xblacklight! Let's build.
Read more: factory.com/news/humanlayer
Ian Butler retweeted
Dear founders,
Go through communication training for both internal and external comms.
Signed me.
Even before AI I thought it was more important for technical teams to act decisively and cohesively versus finding the perfect technical solution.
I think historically there have been very few technical decisions on teams that translated to devastating consequences, instead they translated to some degree of sub optimality that was either very hard for us as humans to really tell or affected scales that most businesses never see and at that point congrats anyway.
As long as the business outcomes were rolling it actually mattered very little what the choice was as long as it was executed in a timely fashion.
You could even paper over poor decisions by putting specific people on dealing with any of those problems as they arose. (Not arguing for this, just that it was an easy bandaid)
At the end of the day most decisions were probably something like +/- 10% over/under the average choice someone could make.
I think many people are having a hard time coming to terms with this fact now that AI lays it pretty bare for everyone to see.
