Watched 'You Can See Evrything' yesterday. All I could think about was Zuck turning down Yahoo's billion because he'd just build another social network, and the Bumfights guy showing up on Dr. Phil.
23
things my personal agent did this month: got my refunds, negotiated my bills, watched my kid's grades, tracked my health, coordinated plans with my wife's agent. user experience is great. there are security growing pains, but as long as they tighten it up, we're good @noahrshinn
1
69
instinct and muse shipped phone calls the same week. parity moves fast now. the fight that matters: which agent gets you the outcome
1
59
hire somebody to do your HVAC. if he's a random guy off a flyer and the house blows up, that's on you. pay a licensed, vetted contractor and the contractor shares the blame - you paid them to get you to the outcome. and if the contractor leaves the gas line unconnected, it's not just your house - the whole neighborhood goes.
AI liability should work the same way. four tiers:
1. you use a frontier lab's model (Anthropic, OpenAI): the lab carries liability. if they claim a $30T TAM, they can fund real safety infrastructure - KYC on customers, logging, actual protections.
2. open source on a pure host (Hugging Face type) that takes no responsibility: to be fair, their terms already say the uploader owns it. then the business model should match - price like infrastructure, not like a lab. you can't disclaim safety and charge lab prices.
3. open source hosted with lab-like diligence: logging, KYC, regular testing for bad requests, flagging. they sell vetting, so they share the liability.
4. open source self-hosted on your own infra: fully on you.
@chamath @bgurley @8teAPi is this the right framework for thinking?
1
98
Everyone's building agents that help you buy. The mirror business is agents that cancel, renegotiate, and save for you - paid out of the savings they create. Sell-side gets the attention, buy-side gets the loyalty. Who's building that? @tobi
55
Feature idea for personal AI assistants: a leaderboard. Who delegates the most, longest streaks, most outcomes closed. Every sticky consumer product has a scoreboard - why doesn't your assistant?
48
Can someone explain to me why OpenAI and Anthropic allow any employee to tweet anything they want.
Apple, Google, Facebook, Nvidia, Microsoft, Tesla, Uber and Airbnb all have obvious and granular rules about e
employees speaking for the company.
Which is to say, employees should never say anything publicly unless it's cleared with comms first
which isn't intended to censor anyone, just to make sure the company is aligned and marching in unison
I could never imagine tweeting the stuff that OpenAI and Anthropic employees are tweeting without checking with the CEO and founders first!
So are these P-doomers clearing their tweets with the CEO and comms?
I don't know if AGI is here yet, but using Instinct I feel AGA - Artificial General Assistant - is definitely here.
54
Been using instinct ai assistant for a few days and it clicked: I'm exactly the target audience. First early-stage product I've used that actually does what it's supposed to do, across a stack of real use cases. Super delighted. And yes, it helped write this post. I just approved it.
73
Most AI product roadmaps start with the model. Start with the decision: What must the user decide? What evidence would change it? Where is a wrong answer expensive? When must a human take over? Then choose the model, tools, and workflow. AI is a product system, not a checkbox.
1
37
Alright, Seahawks got it done tonight. Looking forward to the Niners tomorrow - hopefully they go 1-0 too.
1
64
LLMs are like brilliant blind people. They build rich mental models, spot patterns most of us miss, and reason through messy problems. But they still need instruments, feedback, and a human who knows when the map doesn't match the world. Use them as copilots, not drivers.
44
American is very similar to the roman empire, and studying the rise and fall of roman empire will have a lots of lessons for us to navigate this decade.
1
57
LLMs are like brilliant blind people. Deep understanding of the world. Rich mental models. Pattern recognition that puts most of us to shame. Just don't hand them the steering wheel yet!
30

