Anthropic co-founder Jack Clark just sat down for the Oxford 2026 Cosmos Lecture - and laid out three concrete AI predictions on the record, with specific timelines and his name attached
"60% chance that by end of 2028 an AI system can be told to build a better version of itself and just do it. that is the line. once we cross it, the curve is no longer ours"
"within 12 months an AI will help win a Nobel Prize. not assist research from the sidelines. be a named contributor to the discovery"
"within 18 months there will be companies entirely run by AI generating real millions in revenue. one human owner, zero human employees, real customers paying real money"
watch the full talk, then save the setup below. ↓
Figure AI CEO Brett Adcock just sat down for a long-form interview - and walked through how his humanoid robot just pulled an 8-hour autonomous shift inside a real factory, the same shift you would have given to a human three years ago
"this is the year humanoids cross from demo to deployment. we just ran a real 8-hour shift with zero human intervention. that line is the entire industry's inflection point"
"the labor market is short 8 million workers in physical jobs in the US alone. we are not replacing humans. we are filling roles nobody wants to do anymore"
"the home robot is real and it is closer than people think. once we have one humanoid in every BMW factory we have the data flywheel to put one in every home by 2030"
watch the full talk, then save the setup below. ↓
ex-OpenAI CTO Mira Murati just sat down for a long-form interview - and laid out why her $50B startup is about to make frontier AI customizable by anyone, not just the 6 labs that own it today
"the gap between a frontier model and what you actually need is 90 percent of the value. tinker closes that gap. that is the entire bet"
"AI in 2025 was a few giants building one model for everyone. AI in 2026 is everyone fine-tuning their own. that is a thousand times more interesting"
"we raised $2 billion before shipping a product because the investors saw what was coming. we are talking $50 billion now and we have one product out. the market is pricing in what the labs will not say"
watch the full talk, then save the setup below. ↓
Microsoft AI CEO just sat down for a long-form interview - and laid out the exact 12 to 18 month window before every white-collar job is automated
"lawyers, accountants, marketers, project managers. anyone sitting at a computer all day. AI will do those tasks at human level in 18 months. that is the prediction with my name on it"
"software engineers are already there. most of the code is written by AI. their job has shifted to architecture, debugging, and judgment. that is the model for what happens to every other office job next"
"the people who survive this are not the ones with the best skills. they are the ones who become the operator of the AI doing the work"
watch the full talk, then save the setup below. ↓
Cursor CEO Michael Truell just opened Compile 26 and walked through how his $9.9B AI coding tool is changing what an engineer actually does day to day - $300M ARR, used by every senior engineer you respect, and now shipping agents that finish entire features unsupervised
"the engineer who used to write the code now reviews the code. the writing is the AI's job. the judgment is yours"
"we are seeing teams ship in 3 days what used to take 3 weeks. the bottleneck moved from typing to taste"
"the agent does not need you to tell it what to type anymore. it needs you to tell it what to build. that is the entire skill shift"
watch the full keynote, then save the setup below. ↓
Anthropic's Boris Cherny just sat down at AI Ascent and said the quiet part out loud - prompting is dead, you should be writing loops, and Claude already authors 80%+ of all merged production code inside Anthropic
"stop prompting your agents. start designing loops that prompt your agents for you. that is the entire shift"
"our engineers ship 8x more code per day than they did a year ago. it is not because they got smarter. it is because the loop runs while they sleep"
"by May 2026 Claude was writing over 80 percent of the code that merged into our production codebase. we are the canary"
watch the full talk, then save the setup below. ↓
YC CEO Garry Tan just sat down for a long-form interview - 4000+ startups funded, $600B in YC portfolio value, early on Coinbase and Instacart - and said the era of 50-person seed-stage startups is over
"the next 100 unicorns will be built by teams of 2 to 5 people. it is already happening inside the current YC batch"
"i used to need to see 10 hires to believe a startup. now i look at the 2 founders and the agents they orchestrate. that is the whole team"
"solo founders used to be a red flag. in the AI era it is the optimal configuration"
watch the full podcast, then save the setup below. ↓
OpenAI board chair & ex-Salesforce co-CEO Bret Taylor just sat down for a long-form interview - and said the subscription software industry is dead, the per-seat pricing era is over, and the next $100B in software value gets paid per outcome
"per-seat pricing is a relic of the human era. agents do not have seats. the model is outcome-based or it does not work"
"the SaaS industry was built on a 25 year assumption that you sell licenses to people. agents break that on day 1"
"i ran salesforce, i sit on the OpenAI board, i built sierra. every single one of these is telling me the same thing. the old model is over"
watch the full podcast, then save the setup below. ↓
$20B AI CEO Aravind Srinivas just sat down for a long-form interview - 3 years from research scientist to $20 billion company, ex-OpenAI / ex-Google / ex-DeepMind, building the thing that is replacing Google search - and named the ONLY trait that matters in the AI era
"the secret is sleeping with the fear that a competitor will steal your idea tomorrow. that is the trait"
"fear of failure is the stupidest thing holding people back. i have nothing to lose. that is the entire mindset"
"AI has a glorious future because people don't enjoy their jobs. nobody is mourning what gets automated"
watch the full podcast, then save the setup below. ↓
a16z legend Marc Andreessen sat down with Replit CEO Amjad Masad for a long-form conversation - and laid out exactly what happens to software engineers when "good enough AI" arrives before AGI does
"you do not need AGI to kill the old software industry. you just need good-enough AI. and we are already there"
"replit is building toward 1 billion developers. not 1 billion CS grads. 1 billion people who can ship a product without ever opening an IDE"
"the boilerplate is dead. if your job is writing it, your job is too. that is the entire signal"
watch the full podcast, then save the setup below. ↓
Anthropic CEO Dario Amodei just sat down for a long-form interview and called the top of the AI exponential - and laid out exactly which jobs disappear next
"AI could write 90% of code in 6 months. that is not a prediction, that is current trajectory"
"50% of entry-level white collar jobs could disappear in the next 1 to 5 years"
"we are near the end of the exponential. what comes after looks like a country of geniuses in a data center"
watch the full podcast, then save the setup below. ↓
ex-OpenAI cofounder & ex-Director of AI at Tesla Andrej Karpathy sat down for a 2.5 hour conversation - and quietly popped the AGI bubble while explaining what actually comes next for engineers
"AGI is still a decade away. the models are not there"
"we are summoning ghosts, not building animals. that is the entire framing problem"
"it is not the year of agents, it is the decade of agents"
watch the full podcast, then save the setup below. ↓
solo programmer Pieter Levels just laid out his entire stack in a 3-hour conversation - 12 startups, $200k/month combined revenue, 1 PHP file, no employees, no VC - and he ships every product from a laptop in a different city every week
"i use PHP and jQuery. that's the whole stack. people get upset about this"
"i made $140,000 last month. i don't have employees. i don't want any"
"i built 12 startups in 12 months. most of them failed. that is the entire trick"
watch the full podcast, then save the setup below. ↓
2 people. 90 days. $1M business.
The founder of a $1.3B AI company just said it on camera.
Alex Mashrabov ran AI at Snap. Then he walked out, built Higgsfield, hit $200M ARR, and closed a $130M Series A at a $1.3B valuation. He is one of the few people on the planet running a successful AI company telling other founders to build smaller and faster.
He explains:
- The 2-role split that runs an entire $1M business
- Which AI tools collapse what used to be 5 hires into 1 subscription
- The 90-day product cycle that replaces 18-month roadmaps
- Why most VC-funded startups in 2026 are already structurally too slow to win
- The category his next bet is sitting in (it is not the obvious one)
Bookmark it.
The guy who sold Instagram to Facebook for $1B is now the product chief at Anthropic. And he just laid out exactly which jobs AI starts eating in 2026.
Mike Krieger built Instagram, walked away with a billion, and is now sitting at the steering wheel of one of the two frontier AI companies on earth. He just dropped a long-form breakdown of what is coming for white-collar work next year.
He explains:
- Why coding agents quietly became the breakout AI use case of 2025
- The exact moment enterprises stop using AI as a chatbot and start handing it real workloads
- Which $200K-$400K knowledge jobs Claude is already absorbing inside Fortune 500s
- How to build products for AI capabilities that do not yet exist (the move he is making at Anthropic)
Bookmark it.
The guy who sold Instagram to Facebook for $1B is now the product chief at Anthropic. And he just laid out exactly which jobs AI starts eating in 2026.
Mike Krieger built Instagram, walked away with a billion, and is now sitting at the steering wheel of one of the two frontier AI companies on earth. He just dropped a long-form breakdown of what is coming for white-collar work next year.
He explains:
- Why coding agents quietly became the breakout AI use case of 2025
- The exact moment enterprises stop using AI as a chatbot and start handing it real workloads
- Which $200K-$400K knowledge jobs Claude is already absorbing inside Fortune 500s
- How to build products for AI capabilities that do not yet exist (the move he is making at Anthropic)
Bookmark it.
The CEO of Brex runs his entire company through AI agents. Your email, your calendar, your inbox triage. He automated all of it.
Pedro Franceschi co-founded Brex at 22, scaled it to billions in valuation, and now operates as one of the first CEOs with a full AI agent stack doing his job around him.
He explains:
- The exact AI agents reading his emails, Slack, and running recruiting
- Which CEO tasks AI now handles better than any chief of staff
- Why the next generation of operators will not need 30 reports, just a few agents
- What this means for everyone earning $200K - $500K doing "knowledge work"
If your job is mostly typing, scheduling, and replying, your salary is already on a countdown.
Bookmark it.
The CEO of Brex runs his entire company through AI agents. Your email, your calendar, your inbox triage. He automated all of it.
Pedro Franceschi co-founded Brex at 22, scaled it to billions in valuation, and now operates as one of the first CEOs with a full AI agent stack doing his job around him.
He explains:
- The exact AI agents reading his emails, Slack, and running recruiting
- Which CEO tasks AI now handles better than any chief of staff
- Why the next generation of operators will not need 30 reports, just a few agents
- What this means for everyone earning $200K - $500K doing "knowledge work"
If your job is mostly typing, scheduling, and replying, your salary is already on a countdown.
Bookmark it.
$1M per month. Solo. 5 products.
No employees. No VC.
Tibo Louis-Lucas is doing what every indie hacker says is impossible in 2026.
He just sat down to walk through how it actually works.
- Which products he runs in parallel.
- How AI lets one person operate what used to need a 20-person team.
- Where the actual revenue is coming from inside each product.
This is the conversation that names every excuse most engineers use to stay in their cubicle, and burns them one by one.
Bookmark it.
$1M per month. Solo. 5 products.
No employees. No VC.
Tibo Louis-Lucas is doing what every indie hacker says is impossible in 2026.
He just sat down to walk through how it actually works.
- Which products he runs in parallel.
- How AI lets one person operate what used to need a 20-person team.
- Where the actual revenue is coming from inside each product.
This is the conversation that names every excuse most engineers use to stay in their cubicle, and burns them one by one.
Bookmark it.
Parker Conrad got pushed out of his first company. Came back and built Rippling into a $13B machine that is quietly eating every AI SaaS point solution on the market.
He sat down for a long-form conversation on how to build the next category-defining company in the AI era.
He explains:
- Why the "one tool, one problem" model is dying.
- How AI agents inside one platform are killing 50 different SaaS startups at once.
- The bets that scale and the ones that look smart but cap out.
If you have wondered which kind of AI company actually compounds in 2026 vs which ones get crushed by the next funding round, the answer is in here.
It is a working playbook from someone who is winning the AI consolidation game right now.
Bookmark it.
Parker Conrad got pushed out of his first company. Came back and built Rippling into a $13B machine that is quietly eating every AI SaaS point solution on the market.
He sat down for a long-form conversation on how to build the next category-defining company in the AI era.
He explains:
- Why the "one tool, one problem" model is dying.
- How AI agents inside one platform are killing 50 different SaaS startups at once.
- The bets that scale and the ones that look smart but cap out.
If you have wondered which kind of AI company actually compounds in 2026 vs which ones get crushed by the next funding round, the answer is in here.
It is a working playbook from someone who is winning the AI consolidation game right now.
Bookmark it.
$1.5B AI company. 3 years. One mindset shift most engineers will never make.
Chris Pedregal built it from zero to a $1.5 billion valuation in 36 months. He just sat down to explain the one shift in thinking that separated him from every other founder swinging at the same market in 2026.
What he gets into:
- Why the engineers with the best technical skills lose to founders with worse code
- The specific mindset move that took him from "another AI app" to a category leader
- How to spot the AI bet that has 100x upside vs the one that caps at $5M ARR
- The exact decision in year 1 that put him on the path to $1.5B
Whether you are an employee in a stable job or a business owner trying to scale, this is the conversation that names what you are missing.
Bookmark it.