Mike

@carlvellotti Cc

Every PM should be using Claude Code. So I built a HUGE course for you to learn Claude Code... IN Claude Code! 🔹 Complete guide 🔹 Make PRDs, analyze data, create decks Soon, I'll sell it for $149. For the next 24h: FREE! Follow + RT + comment "CC" & I'll DM it.

@RyanSAdams Basically the @thinkagents premise

I'd love to connect AI agents to my files, email, trading accounts, chat groups - incredibly powerful But i can't do it until i have data privacy and control imagine the data breaches imagine the loss of control imagine a hack - you could be utterly destroyed What I want is turnkey hardware that runs local models that the leading AI agent tools can treat as a cloud - same UX, but privately without compromise Something like @__tinygrad__ Tinybox meets @Ledger There's got to be $10 billion opportunity for the startup who makes this real Because what's the alternative. The AI companies own us? Is it possible to make a fully encrypted, bulletproof private AI cloud.

@nateliason With your track record on tweets like this over the past 10-15 years, you could shill me just about anything and I’d blindly buy/use/adopt it at this point!

I can't believe how much better vibe coding is with @laravelphp. Even with a more complex setup like a Vue frontend. Claude just one shots huge requests without errors. It's wild to see.

@thinkagents Definitely a distribution play. Selling glasses will be different than Quest headsets. "The arrangement gives Meta the advantage of having more detailed manufacturing knowledge and global distribution networks, fundamental to turning its smart glasses into mass-market products."

@thinkagents "Meta acquired just under 3% of the Ray-Ban maker, a stake worth around €3 billion at market prices, and is considering further investment to build the stake to around 5% over time."

@businessbarista Ai

The best AI tools & tactics are stuck in private chats and buried Slack threads. So I'm spinning up a few WhatsApp groups for folks who are actively using AI in their day-to-day. First is an AI group for ceos/founders. Second is an AI group for engineers & engineering leaders. Third is an AI group for marketers & marketing leaders. No sales pitches. Just smart people sharing how they’re using AI to move faster, do more, and stay ahead. If you want to join one of these groups, reply with "ai" and I’ll DM you an invite.

@mikeyanderson Ah yes, so more like “bed preparation”. The soil is clear and loose — it’s time to feed!

@MJC2232 I think I view “aligning energy” as something like gardening. :)

Should we do a Think virtual meet up? Great post below on being intentional about creating strong community. I want to remove the 4th wall as much as possible and bring the community in on the process. It’s about us aligning our energy behind owning our intelligence. That network is gonna grow 🌱 -> 🌳

@thinkagents Couldn't figure it out, unfortunately

@MJC2232 We’ll look into this 🫡

@thinkagents Thanks! I got an error in OpenSea but am guessing that's user error. I'll figure it out. Excited for this project!

@MJC2232 You can send it 🫡

@jessebryan @alexisohanian @RobinHoodNYC @beyond12 @thinkagents @rendernetwork @MeebitsNFTs What LLM is underneath this?

@alexisohanian @RobinHoodNYC @beyond12 Think you will dig what we are building over @thinkagents open source agent platform you custody the data. Just announced our partnership w @rendernetwork & @MeebitsNFTs Here is a lil demo

Big news: the future of AI for good is here💥 As a board member at @robinhoodnyc, I’m proud to announce the 3 finalists of our $1M AI Poverty Challenge: 📝 @beyond12 – AI coaching for first-gen college students 💻 CodePath.org – EdTech unlocking tech careers 🏠 @LegalAidNYC – AI streamlining housing justice These innovators are proving AI can fight poverty—and change lives. Thank you to our Challenge partners—GitLab Foundation, @BezosFoundation, and Deloitte. Stay tuned for what’s next!

@jessebryan Yes please!

Starting a SOULS group who wants in?

@FFVV1211 @thinkagents Interesting project! Any idea why you would want to own/mint multiple agents other than for speculative reasons? I would hope your agent could handle more than one workflow.

Been seeing a few ppl talk abt @thinkagents, seems pretty cool Here’s what I’ve learned: • A protocol linking open-source AI agents, tools & models. • Enables composable, user-owned intelligence. • Decentralized to protect free thought & innovation. • Built for developers, creators & ecosystems. THINK AI Agents: • Adaptive: Gain skills & attributes across platforms. • Interoperable: Carry personality, data & assets seamlessly. • User-Controlled: You decide how agents evolve & interact. • Protected: Innovate under $THINK’s patent shield. Their NFT mint is coming up on the 22nd and I’m excited, this is a crossbreed between NFTs + Coins + Art, let’s see what happens🤝

@punk9059 @GuidoDisalle I don’t think most people realize how dire the debt situation is - we have to hope it can still be salvaged

@GuidoDisalle He’s taking a lot of damage to achieve that end result.

Just chatted w a friend about whether these tariffs are serious or whether they will get removed before long. Big question is why is Trump doing them. I have heard six potential explanations: 1) Long-term goal of on-shoring manufacturing. 2) Increase revenue via tariffs to pay off debt and/or remove taxes. 3) Power and revenge. 4) Negotiating tactic to get certain concessions. 5) Hurt the economy so 10-year rates go lower and debt can be refinanced. 6) FAFO What do you think are the real objectives here? Do you think they will go through?

@lamps_apple @OfficialLoganK @cursor_ai Would love to hear more about what this does

@OfficialLoganK @cursor_ai 2.5 pro built my entire app and deep research ability 🔥🔥

Gemini 2.5 Pro, available to all users in @cursor_ai! It’s time to build 🚢

@david_perell As someone who has read your stuff for years, this post really feels like it was written by AI. Not sure if that matters to you or not but figured the feedback might be useful.

The education system is broken. Costs are rising. Outcomes are falling. Kids are bored, parents are frustrated, teachers are over-worked and under-appreciated. The Wall Street Journal just profiled the most promising solution I've seen. Their results are so good that you're not going to believe me when I share them with you. But by the end of this, you'll understand what they're doing differently. The results: • Alpha students are top 1% in the USA in every class and every subject (except for one: 5th grade math, where they're “only” in the 93rd percentile). • Alpha kids are learning 2.5x faster than the average American student. • They crush standardized testing too. Only 2% of Texas students score above 90% on the Texas STAAR test. Every Alpha student scores above 90% and the majority of their students are over a grade level ahead. • Alpha has more kids who score 100% on the Texas STAAR test than one entire school district of 100,000 people (Alpha has ~250 students). And here's the kicker: Alpha School students only spend two hours on academics per day — just two hours from K-8 and three hours in high school. The time they save gives kids back the rest of their day to learn life skills like socialization and public speaking instead of twiddling their thumbs during soul-crushing lectures or waiting for the bell to finally ring. Sounds too good to be true, I know. So let me break down what they're doing differently. For starters, the founders simply took learning science seriously. They reimagined how a school could function instead of following the standard school playbook. The research showed them that maximizing learning outcomes means building something that looks nothing like our existing school system—which has basically stayed the same for 200 years. Here are three things they do differently: 1) Mastery Learning Knowledge is like a Jenga tower. The higher blocks depend on the stability of the lower ones. If you pull a crucial block at the bottom, like knowing your multiplication tables, the whole structure can collapse. For example, good luck doing algebra if you don't know your multiplication tables by heart. The problem with traditional schools is that because every student in a grade has to move at the same pace, whether they know the material or not, kids end up with holes in their knowledge. This conveyor belt approach sets them back forever. Not knowing multiplication won't just stop you from crushing algebra. It'll stop you from crushing statistics and calculus, and entire career paths once you’re an adult. Mastery learning simply means that you don't move onto the next level until you know the material from the previous one. It's the opposite of how schools work right now. In traditional learning, time is fixed (everyone gets exactly one school year for algebra) while learning is variable (some kids get A's, others get C's). In mastery learning, it's flipped: learning is fixed (everyone masters the material) while time is variable (some finish in three weeks, others in three months). 2) Personalized learning plans If you want to learn fast, you need challenges that are tailored to your skill level. If the material is too easy, you’ll be bored; if it’s too difficult, you’ll get frustrated and quit. What you need is the sweet spot of content that’s challenging but doable. It’s the Goldilocks Principle for school: not too hot, not too cold. Learning researchers call this the Zone of Proximal Development. It’s why 1-on-1 tutoring works so well. The problem with traditional lectures is that teachers need to teach to the average, which means that most students don’t get what they need. Visit a struggling charter school and you'll see this in action. Ask a 6th grade teacher what they're teaching and they'll say: "Sixth grade material. That's my job." Meanwhile, half the class can barely read at a 2nd grade level. Teachers know this, but because going back to the basics could cost them their job, they plow ahead, which leaves students behind. The cost of these perverse incentives is showing up in the declining test scores we’re seeing across America. 3) Change what teachers do Picture the typical classroom: one teacher at the front, 25 kids of wildly different abilities, and 45 minutes to get through the material before the bell rings. Meanwhile, teachers are stressed during the day and overworked at night from all the papers they have to grade and lesson plans they have to write. And for what? For 40 years, research has shown that passive lectures are just about the worst way to learn anything. It’s an established fact. But our education system is so fossilized that it’s incapable of adapting to new knowledge and the better way of doing things that we’ve already discovered. So what does Alpha do? They’ve transformed teachers into coaches and motivators. The apps and AI handle the information delivery and customize it for each student, leaving teachers to do what they do best (and signed up for in the first place): connect, inspire, mentor, and motivate students. The bottleneck isn’t information anymore. It’s student motivation. You can have the world’s greatest curriculum, but if the student doesn’t give a damn, they ain’t gonna learn squat. Everybody talks about EdTech, but it’s only part of the solution. Motivation plays a far more important role, which is why the people I know at Alpha are always asking: “How can we get kids to love school?” — — The Science Behind Alpha's Approach These strategies are inspired by Bloom’s Two-Sigma Model, which is arguably the most important education finding of the past century. When researchers combined mastery learning with 1-on-1 tutoring, 98% of the tutored students outperformed the average student from a traditional classroom. Here’s the key point: Mastery learning, paired with 1-on-1 tutoring, is better for the very best students and the very worst ones. 1-on-1 tutoring isn’t some futuristic approach. Descartes, Feynman, and John Stuart Mill all learned through tutoring. The problem with tutoring, of course, is that it’s always been reserved for the elite. It’s too expensive and there simply aren’t enough master tutors to go around. Enter modern technology: AI and learning apps have changed what’s possible. Every student can have a personalized learning plan now. Digital tutors never get tired or cranky, and they’ll certainly never roll their eyes and say: “Come on, kid… we’ve gone over these fractions twelve times already!” They’ll work all night with a kid who wants to memorize their fractions or learn the capitals of every country in the world. With each question, the software develops a laser-precise understanding of what concepts each kid knows / doesn’t know. Once again, that leaves the teachers to focus on the human side of learning: motivating students and helping them when they get stuck. — — Science fiction author William Gibson once said: “The future is already here; it’s just not evenly distributed yet.” And that’s exactly what Alpha School represents: a future of education that already exists, and not some pipe dream vision of a distant tomorrow. I know because I’ve witnessed it first-hand. While running my own adult writing program, one of our paid editors raved about a participant’s killer writing skills. “Peyton has serious talent,” she wrote. That talented writer, Peyton, was a 15 year-old Alpha student at the time who was out-writing many of the adults in the program. I’m obviously fired up about what Alpha is doing, but if the idea of kids learning through apps and AI makes you uncomfortable, I get it. But unlike our current one-size-fits-all system that’s failing millions of kids, Alpha isn’t trying to be the universal solution to education. Their message goes something like this: “We’ve built something that works very well but looks nothing like the school you went to. We’re not trying to be conventional. We’re trying to be effective, and that means doing unconventional things. If that resonates with you, great. If not, that's okay too.” There’s something here. The data speaks for itself. The most untapped resource on the planet is human potential, and transforming the education system is just about the best way to unleash it.

@nateliason Do it. Pure optionality. Minimal downside and enormous upside.

I feel like I should just try to show up at this point

@mukulagarwalXR @petergyang Helpful - thank you!

Sure. Deep Research is a research agent system created by OpenAI. It can take a research question and search the web and provide a highly detailed answer. Please create a prompt for this system based on the topic below that is: - Well reasoned - Highly Detailed - Clearly structured - Achieves a precise goal Topic: Put the topic that you want to research here

I finally caved and paid $200 to try OpenAI's Deep Research and Operator. What have been your most useful use cases for either one so far?