Israel Adelaja

I'm Israel, a 19 y/o Australian founder building in robotics / physical AI. I’ll be in San Francisco this week. A bit about me: - Coding since 9 - Been building tech businesses since 14 - Prev built two profitable AI startups If you work at a physical AI lab, or an infra or deployment company, I would love to visit your office or have a coffee chat. I want to hear what it takes to get robots working in the real world, and what’s holding us back. I'm happy to share what I've learnt as well. I’ll also be at Long Horizon and a few other events. My DMs are open!

“we’re building autonomous factories” looks closer.. it’s an ERP software that makes LLM calls

san francisco, day 7 last full day in san francisco. rode a @Waymo for the first time. i've known about it for a while but sitting in one and watching it drive itself truly amazed me. it took waymo more than a decade to go from an internal project inside google to fully driverless rides. felt fitting to see probably the best example of "physical AI" deployed at scale on my last day. also took @Caltrain for the first time to mountain view for a robotics event. mountain view is a lot quieter than SF. would be a nice place to live. then ended the day with some @innoutburger. growing up in australia, it felt like entering the set of an american TV show. i fly back to sydney tomorrow. this trip has definitely been a big inflection point for me. i met so many incredible people, and i’m leaving with a new perspective on the next steps for my startup. i’m ready to hit the ground running back home. i’m writing a longer piece on my time in the valley. more soon!

sometimes i meet someone i've only seen on twitter and i have to remind myself im in SF

san francisco, day 5 attended a fireside chat at founders inc with the cofounder of physical intelligence (Quan Vuong). something that stood out to me from the talk is that physical intelligence don’t actually know which type of data modality (egocentric, UMI, teleop, etc.) will win long term, everything is still an experiment. i also had the chance to visit @MissionRobotBay where some researchers/companies talked about what it takes to deploy robots in the real world. i enjoyed Brian Tsui's talk on robotic control via LLM agents (hot topic now given the recent advancements with GPT 6 Astra). also met some cool robotics people, including @Scav, who gave me advice on what i'm building. talking to robotics people in SF there is a growing concern that VLAs and WAMs alone will not get us to reliable deployments. these robots need to be 99.9%+ reliable and even if you can get there with imitation learning it will take you months of teleop + intervention data just for one task. as a result some deployment companies are pivoting to selling egocentric data with the hope that the physical AI labs will figure something out soon. i don't believe that just collecting more data will solve the problem (at least just by itself). robotics is different to language models; there are many factors to take into account in the physical world. we need to think about how we can build systems that deploy intelligence into the physical world with precision and reliability. this is the problem i'm passionate about.

san francisco, day 4 spent more time at the @fdotinc campus. i've been learning a ton speaking to the founders there. in the evening i went to a small robotics event. while there someone i’d just met gave me ~20 mins of advice on my startup. it made me reflect on the "pay it forward" mentality in SF. when i meet someone new there is a culture here of doing whatever is possible to help (even if it's little), and with no expectation in return. like an intro or advice on what you're building. i feel it's a big part of why things move so quickly here.

ive noticed this a lot listening to speakers and talking to people at long horizon. its a bit concerning to me because it looks like just copying the language model framework will not work for robotics.

san francisco, day 3 first day of the long horizon physical AI summit. i spoke with people across robotics and physical AI including: @chris_j_paxton (ai @ agility robotics), @Grace_JBrown (founder & ceo @ andromeda robotics), @AndyChenML (head of robotics @ runwayML) and more. here are some things that i learnt and am still thinking about: - from the people i spoke to, the dominant recipe for robot deployments is imitation learning with dagger style intervention data. - the bitter lesson also applies to physical AI. direct control with general models will become more widely adopted, but there is still room for specialized robot models (like dream 1 that just launched) - classical robotics likely still has a part to play in deployments because there are some tasks where it makes more sense than a trained policy. an interesting idea is using an agent harness for direct control where it creates "skills" for itself, which are deterministic code it can reuse over time. - pretty much every company in physical AI (even the big ones) is still experimenting. nobody has this figured out, which is what makes it exciting. i also had the (quite spontaneous) opportunity to visit the @fdotinc campus and speak with founders like @IsaacSin12 (building @ makermods), @davidim (ceo @ sume) and more. they are all building really cool stuff :)

finally at fort mason :D

san francisco, day 2 i spent the first half of the day between two hackathons on the same street. the original plan was to go to one of them (an AWS hackathon) but it was only when i got there that i realised the other hackathon was happening on the same street. so in my bid to network with as many people as possible, i jumped between both instead of sticking to one 😅 i did meet someone at the AWS hackathon who told me that the south park area has a very high density of hardware/robotics startups. i am thinking of just walking around (or just hanging out at a cafe in the area) and seeing if i can serendipitously meet anyone that way, but i am only here for a week so my schedule may not allow, we'll see. anyway, after (unsurprisingly) not winning a prize in any of these hackathons, i headed over to my first official "robotics" event since being in SF. naturally this one felt like the most productive event thus far, so much so that i was able to book an in office meeting with a pretty big robotics startup in SF (likely more in this soon).

SF is literally tech twitter in real life. im just walking down the street and the people behind me are talking about “openrouter” and “jev” 😭

just saw a poster in SF saying “speed is the new moat”. i love this city man.

san francisco, day 1 got to SFO airport in the morning, checked into my accommodation and headed straight to a @spc event. it has nothing to do with robotics (was about biotech) but it didn't rlly matter to me. one of my main goals this week is to make as many connections as possible, and going to events where high signal people congregate can open many unexpected doors. during the event i had the opportunity to talk to @mjacobstein and some of spc's super talented portfolio founders. i've developed a sort of "formula" for creating USEFUL conversations during events (esp. with important people): - don't be scared, they are normal humans like u, and worst that can happen is they discretely try to run away from u (has happened to me a couple times haha) - introduce yourself and why u are worth talking to, don't waffle, only important stuff - clear ask (this is the most important otherwise it's a waste of time for both of u). the more practise you get the more you stop caring about what people think and just start creating these conversations naturally and confidently. once u get over the sometimes awkward initial greeting, the convo is smooth sailing from there. will be applying this formula throughout the week (also need to remember to take photos with people i talk to haha).

man it’s so trippy driving around SF and seeing billboards of companies ive only seen on X. im so excited to be in close proximity with the elites of the valley.

im in SF! first time in the states, the land of opportunity and the land of the free. God bless the United States of America 🫡🇺🇸

on my flight to SF rn, haven’t even taken off yet and already heard the term “YC startup” being thrown around 😭