Absolutely blown away by the turnout at @FactoryAI tonight!! Huge thank you to the incomparable @dee_bosa @cjc @NancyZWang for joining us and to our all star @theashleyjepson ✨ for hosting More to come! 🙌 @francesca_lab @sonofalli @eliza_bara @maddiehfaulkner et al

Ten of my friends are gonna ask to hold my iPhone Duo on day one. If your app is on it and it's fun, they download it right there so it's ready when their own Duo shows up. So put your app out there now, even if you don't have a Duo yet.

I think it feels natural to use the iPhone Duo this way. If I hold it with the cameras at the bottom, my hands cover them. So I keep the cameras up and the little pinky in the bottom so it doesn't slide on down. God forbid you'd wanna fold a display.

Chat told me you can only do speakerphone when the iPhone Duo is open. Apple says opening it during a call moves the audio to the speaker. But if you got AirPods, bro. Also, who answers phones in 2026? All of my calls get routed through their own spam server.

If Terra can figure out how to do secure auth for websites it would be top tier. I tried Grok Bot vault yesterday and felt just like Apple passwords and the agent wouldn’t use keys that are outside of the domain it is assigned to (gmail, etc). Hint: just copy roadmap from poteto (:

I don’t talk about my background on here often, so I figured I’d give a breakdown for some context on me! I’m incredibly lucky to have grown up in a very entrepreneurial family. I have a lot of fond memories of helping out with the family businesses from a very young age! When I was a teenager, we started ramping up our family lifestyle and travel blog. We traveled 150+ days a year from 2013 to 2019, filmed videos, wrote a lot, and worked with a ton of incredible companies. It was amazing (exhausting too, but so amazing). From 2017-ish to 2022, we ran a program where we taught other people to blog! This was my first foray into online education and community building. I spent pretty much 24/7 teaching and advising students in our Facebook group (remember the good old days of Facebook groups?). It’s incredibly rewarding to have the opportunity to nurture a big community and to help members of that community become really successful. In 2021, my dad "stumbled" into land investing. He started building up a business around it, and then I got interested and decided to join in! I had so much fun working in land that I decided to co-found another land investment company with my sister. We 10xed our investment that year, which I’m still pretty proud of. Then we got a little tired of competing with our own dad, haha, so I rejoined his company and started leading deal evaluation. This was so much fun. Land deals have so many moving pieces, and it’s an absolute blast to put the puzzle together and be able to see the value of a deal clearly. I still absolutely love evaluating deals! I also started mentoring others on land investment through another high-ticket program and community that we started. I still consider one of my favorite accomplishments to be when I gave a piece of advice that ended up adding $50k to the profit total of a student’s first-ever deal. He was sold on investing after that, haha. Last year, my dad and I also expanded into a renewable energy (utility-scale battery energy storage system development) startup called Sunland America. We raised $895k and made some great hires! I’ve taken an advisory position in the company at this point while I build, and it’s still going strong! I’ve built a ton over the past year (hardware, software, everything I can), won five hackathons, hosted events, attended events, consulted for some great companies, and I’m really just seeing where this all takes me. I'm not quite sure what the future holds, honestly. Whatever it is, I can't wait :)

We spent fifty years making proofs impossible to fake. Then we let the coding agent pick what to prove. A research paper this month put a number on it.* AI wrote the formal spec, AI wrote the proof, the verifier accepted both. Then humans read the specs. Close to 60% proved something other than what the author meant. Not wrong proofs. Wrong specs. The verifier was right every time. That's the problem. Left to choose, the agent picks properties that are easy to prove and say nothing about the program, and the checker can't tell a useless theorem from a useful one. I'm calling it vibe verification: the agent writes the spec and the proof, the checker goes green, nobody reads the spec. This applies to Verus, Lean, Bend, TLA+, Dafny, [insert favorite formal language here]. Here's an example of an agent failing at something that feels too simple to fail: Ask for a verified sort. It writes a spec that says "the output is sorted" and proves it. What the spec never says is that the output has the same elements as the input, so a function that returns an empty list satisfies it, and the verifier accepts the proof. Trivially satisfiable specs are the default failure, not an edge case. Here's a formula that actually works. Don't say "formally verify the codebase." Say "show that [property] holds in [component]." A property is a precondition (true going in), a postcondition (true coming out), or an invariant (stays true the whole time). For example, "A session token is never accepted after the logout that revoked it" is a good invariant. The bottleneck is moving from code review to intent review, and the spec is where intent lives. So: read the spec before the code. Write 3 inputs it must accept and 3 it must reject. Treat a model-written spec with utmost paranoia and rely on your judgement as an engineer. Formal verification for software doesn't automatically create trust, it moves it, from the code to the spec. If a human didn't read the spec, you didn't verify anything. You vibe-verified it. Follow me for more on verification and agentic engineering. * Wang, Jiang, Liu et al., "VerusEval: Evaluating LLM-Generated Verus Specifications Beyond Verifier Acceptance", Shenzhen University, arXiv 2610.26464, October 2026

I did 40 plus releases of Ray Transcribes from a boat over two weeks. It all runs from one Cursor Project with a release rule, and the agents test every release in their own cloud environments. Write the rule before you leave. Cursor gave me research credits to push this.