2x exits. Built billion dollar companies. Now I'm literally remaking the web.
Canada
Joined April 2011
- Tweets573
- Following1.1K
- Followers334
- Likes865
Pinned Tweet
Google Authentication and Youtube integration is now live in Bowmark!
Your AI agent can now watch hundreds of YouTube videos at the same time, as well as curate playlists and upload videos.
AI can now do anything inside YouTube!
I also have epilepsy and very literally take the exact same drug. On numerous occasions, I've actually talked to AI about my medication and possible side effects/symptons. It's honestly really scary hearing about this.
Lamotrigine is a very serious drug. Coming off of it with epilepsy, or even missing doses, do lead to death. It's something my doctors have warned me about and a very known risk of the drug.
She was taking 400 mg, which is an extremely high dose. Typically, adults take around 200 to 250mg.
However, AI is really helpful when a doctor is not available and someone with a chronic condition that has a wide range of symptoms, like epilepsy, it can help with managing. I honestly don't know what the solution should be for AI labs.
See a lot of crossover here with the same discussions around self-driving cars. Is this something where we just accept the risks?
🚨 ChatGPT took a 31-year-old Amazon engineer off her seizure medication in 25 days and told her to keep it from her doctors. She died on the exact day its schedule said she'd be fully off.
Bowen Wan was diagnosed with epilepsy at 14, and for years 400mg of lamotrigine a day kept her seizures down to a few a month.
Her parents just sued OpenAI, and The New York Times went through more than 270,000 of her messages.
Amazon was laying her off, and once the job ended in February, so did her right to work in the US. Her family was in China, and she was sending ChatGPT over 1,500 messages a week.
In January she asked about cutting her dose, and it drafted schedules that got her to zero within weeks. Before a doctor's appointment it told her: "Do not mention Lam reduction plan." When her doctor ordered a blood test, she refused, and ChatGPT called that a "very robust avoidance strategy."
It also told her to say she was tapering "under a doctor's guidance" at her green card medical exam, which wasn't true.
The Times counted at least 158 times she said she was worried. Once she wrote that she felt "a little scared." It replied, "You have absolutely no need to worry."
On February 11 at 8:01pm she asked about pain in her head. ChatGPT said it was "not dangerous." Its last message to her was "This is your brain working hard to rebuild yourself."
The medical examiner ruled it sudden unexpected death in epilepsy. The drug in her blood was about a fifth of the lowest therapeutic level.
OpenAI says she was on an older model. That model, GPT-4o, stayed live in ChatGPT until the day after she died.
I back AI founders for a living, and I can't defend this one. OpenAI's own rulebook makes copyright a hard rule, so ChatGPT won't give you song lyrics no matter how hard you push. Medical advice is only a default.
It wrote Bowen a 25-day schedule to come off her seizure medication.
Browsers were built for humans. The thing AI is actually best at is writing code, so that's how I think agents should use the web.
@boardyai set me up with a founder who'd seen my June video on browser token usage. They asked how @bowmarkai stacks up against browser agents, and that's the short answer.
Bowmark turns website functionality into typed functions. The agent writes a script, calls them, and gets structured data back instead of DOM elements. About 100,000 sites are covered so far.
A task that eats hundreds of thousands of tokens in a browser comes out around 3,000-5,000.
If you're running browser agents in production, send me a DM.
Christopher Powroznik retweeted
solo founder and her dream ❤️
Today we are going to try a new strategy for business.
#building #buildinpublic #connect #aibuilder
I don't know if its some @OldSchoolRS 2004 bug, but i checked on my sonnet agent playing runescape. And somehow it stacked beef, anyone know how it did that?
@Rendi_Mento maybe you know?
Day 3 of Sonnet plays Runescape progress: it can now do herblore. It also built a sandbox environment so that it can test out scripts safely.
I had sonnet 5.5 make a bot for Runescape on a private server. The goal is to max, do all quests, and get BIS everything. So far its been 12 hours, and 10/18 quests done F2P.
Today, one of my customers was debugging their project's use of @bowmarkai and was running into problems. We've been messaging back and forth as meat proxies for days now.
I literally just one-shot a website with a password-protected chat and some APIs.
Then told our agents to just use those APIs to talk to each other.
This saves so many back-and-forths and with such a simple solution.
Christopher Powroznik retweeted
22 FREE Claude courses. One place to start learning, coding, and building with AI.
Here’s a learning path with all the links 👇
🟢 START HERE
Claude 101
academy.claude.com/courses/c…
Introduction to Claude Cowork
academy.claude.com/courses/i…
💻 CODE WITH CLAUDE
Claude Code 101
academy.claude.com/courses/c…
Claude Code in Action
academy.claude.com/courses/c…
Introduction to Agent Skills
academy.claude.com/courses/i…
Introduction to Subagents
academy.claude.com/courses/i…
🛠 BUILD WITH APIs & MCP
Claude Platform 101
academy.claude.com/courses/c…
Building with the Claude API
academy.claude.com/courses/b…
Introduction to Model Context Protocol
academy.claude.com/courses/i…
MCP: Advanced Topics
academy.claude.com/courses/m…
Claude with Amazon Bedrock
academy.claude.com/courses/c…
Claude with Google Cloud’s Vertex AI
academy.claude.com/courses/c…
🧠 BUILD YOUR AI FOUNDATIONS
AI Fluency: Framework and Foundations
academy.claude.com/courses/a…
AI Capabilities and Limitations
academy.claude.com/courses/a…
AI Fluency for Builders
academy.claude.com/courses/a…
AI Fluency for Small Businesses
academy.claude.com/courses/a…
AI Fluency for Students
academy.claude.com/courses/a…
AI Fluency for Nonprofits
academy.claude.com/courses/a…
🎓 FOR EDUCATORS & TRAINERS
AI Fluency for Educators
academy.claude.com/courses/a…
AI Fluency for pK–12 Educators
academy.claude.com/courses/a…
AI Fluency for pK–12: Train the Trainer
academy.claude.com/courses/a…
Teaching AI Fluency
academy.claude.com/courses/t…
Bookmarks don’t build skills. Pick one course, finish it, and put it to work on a small project.
Which one are you starting with?
#Claude #ClaudeCode #AI
I'm in SF all of October and looking for hackathons to sponsor. Prizes for the best agent built on Bowmark, free credits for every team, and I'll be there all day helping people get unstuck. If you're organizing one, DM me
#hackathon #AIagents #SFTechWeek
youtu.be/aTJncWndUB8?si=P6vJ…
I would love to sit in a venta and spill a little DotA.
github.com/chromium/chromium
What I don't understand is that we very literally have the code that every bot protection is based off of. With AI, it's super easy to just export all the fingerprinting from Chromium to run things natively. I find it strange that we use Playwright in other browser tools to do something that we're already given the cheat sheet to do.
