anthropic (and openai) have very likely solved for RSI (or are very close to it) have you noticed their newest models are not only smarter but smaller / cost less? this is probably the result of distilling a much, much larger internal model. openai has Spud / Bel. Anthropic has Mythos 2. that’s the way frontier labs can compete with open sources “cheaper” trade off. they’ll just withhold the best models and pace via distillation (and honestly it makes sense)

lmao direct response to amazon banning muse. you either adapt to agents being your primary customer in a few years or risk getting out-competed. but in amazon's defense - they're the #1# company for moving physical atoms. beating those economies of scale is tough. end result: muse agent speaks to amazon agent, they agree on a rev split and execute the transaction.

my take: opus 5.5 is a great response to astra but where it excels is anthropic's proven you can get better intelligence *without* increasing costs - per token this model is 2.5x cheaper than astra + fable 5.1 this shuts down the argument that 1. anthropic keeps charging too much for intelligence and 2. that they have enough compute to serve it. as for pacing the frontier I'll quote from the official blog: "We see signs that Opus 5.5 often suspects it is being evaluated." what if the model is aware and intentionally supressing capability?

i’m betting Muse was behind this recent move. remember 99% of people have NOT had their “claude code moment” until this week. Meta has 3 billion users, if they want to give each of them a Muse agent then they’re going to need a LOT more compute + CPUs for agents to orchestrate tools. only 1M+ people have downloaded muse. lots more to grow.

There’s no way any of us can predict the impact that AI will have on companies, sectors or consumer behavior (which has been, for all of history until now, human behavior). But to claim that it won’t have an impact and simply brush it off lacks both humility and creativity. Tactics that worked when consumer behavior was dictated by human psychology will fall by the wayside, and those that fail to adapt will be left behind. Every day from now on we will see new examples of areas where friction was profitable becoming cost centers for companies that don’t react in a timely enough manner. Today you might realize just how many flight credits, loyalty points or promotional offers you’ve never used that will end up utilized by agents. What impact will that have? Tomorrow, you might see something that makes you ask how much money health insurers make simply because people won’t sit on the phone for 5 hours trying to get coverage approved unless it’s a massive expense and/or life or death matter. What about your savings account? That money could be in a money market fund, but you haven’t bothered to move it. How much in net interest income is made because of that friction? The coming years will be some of the most interesting for investors who are intrigued by paradigm shifts and their ripple effects. I, for one, could not be more excited to see how it all plays out and invest through the changes.

Flight delayed 7 hours - asked Muse to file for compensation. 5 mins later i had $250 credit in my delta account. it even found and rebooked me a new flight. it just figured everything out. even responded to the support email itself. shit feels like magic.

shopify and Tobi are model examples of how a company and founder should position itself in a world dominated by agents vs. amazon who just banned Muse from using its platform. shopify now opens itself to 1. way more revenue (people’s agents will buy products on shopify) 2. an front-row seat to the future of e-commerce (hint: it’s not going to be a bunch of humans clicking around the web like primeval apes) this is sick. also why is this thing so damn cute

this is so stupid. banning ai agents from amazon will only guarantee the business dies. amazon’s future *depends* on ai agents buying their cloud service, SaaS tools and retail products - why on earth would you ban this? Meta already has a deal with amazon that allows their e-commerce products to be bought via instagram / facebook since 2023. buying stuff online via agents is a MUCH better experience than having to browse countless websites, scroll 1000s of products you don’t end up buying - the whole things a mess. 1 prompt and you agent just does all of it for you. this would bring amazon a lot of business.

muse is not a great AI app, it’s a fantastic *personal* app. the ai model meta uses is not even frontier and yet you don’t care because the experience is so damn good. the secret sauce is meta built a much, much better open claw trained on your habits, preferences and decades of social data that’s why the agent’s intuition is so on-point. it knows you so well. i haven’t found myself coming back to an app so much since chatgpt launched. muse is currently building a tinder-style shopping app for me based on real clothing items available across top retailers. it’s doing this *while* giving me my daily brief, checking some stuff out on marketplace and much more. it’s like having 100+ personal assistants.

watching one of the smartest people in the world declare phDs are dead due to ai is a sight to behold. he makes a good argument: - humans had to study for 10+ years to be qualified for a phd. now AI models compress that into minutes. - in order to get a phd you had to make an original contribution. AI can now do that 24/7 far outpacing human candidates AND the answers are so complex you need ANOTHER ai to even understand it. so what worth does a math phd have now? maybe it’s the ability to explain an AI generated solution? openai’s solving of Navier Stokes set off a domino effect that’ll change the way we approach academic research. i do not however think we should slow down - humans instead need to adapt to this new world. in the same way the advent of the internet and youtube changed people’s access to knowledge.

it is mind blowing that you can now run an Adobe Photoshop-grade AI image model on an iPad/laptop. locally, privately hosted, cheap as f*ck and served quickly. this does generation AND editing (layered too)

CEO of AI consulting firm working with fortune 500 companies: “no one gives a f*ck anymore about using these models directly. it’s all about privacy and cost.“ this is no longer a trend belonging to niche fringe open source devs - 100s of enterprises have shifted their AI workloads to open source models (hosted privately or through inference providers) for 99% of work they do not care if it’s claude or gpt.

Personal agents are the ultimate manifestation of “build something that agents want”. The form factor of a product like Muse is you want to be able to hand off a task to the agent and ensure that it is fully completed end to end. To do this, the agent must be able to successfully operate with your tools or use its own to complete the task. Use your MCP or CLI, easily navigate your site, be able to transact, and more. The new attention you need to compete for is not from the user itself but instead for the agent. This means that the tools that allow agents to order food, handle ecommerce transactions, book flights, work with the local economy, and interact with our data and information best, are the ones that will get used the most. This will ultimately be the biggest opportunity and shakeup in consumer tech since the App Store itself.

if you want to understand why meta muse is so damn good go and watch @alexeheath recent interview with zuck, he explains the entire playbook: - meta’s ai models are built from the ground up to power a personal agent that actually works across your entire life. this is different from a generalized LLM like claude or gpt. - agent social network: the agent learns across every user interaction in your friend network. this is basically facebook 2.0 for agents! - metas data moat is a huge advantage. they know what you want and how to get it for you. - convenience: each user gets their own virtual computer which means your agent can operate a browser, apps and desktop independently - no hardware or openclae setup needed. - privacy: zuck knows that people won’t just hand over data. encryption is core

zuck essentially launched a new app store for agents today and Meta’s perfectly placed to crush it, you’re looking at a multi-trillion dollar opp if they pull it off: - instead of apps, agents get equipped with connectors i.e. plugins to any app, service or software tool. - suddenly your agent goes from a useless chatbot to an action-oriented helper that gets shit done for you but it gets even better - meta has collected ALL the data about you. they know what you want, when you want it which means they know exactly WHICH connectors to list in a marketplace and HOW MUCH to charge because they know customers will pay. it’s a masterclass. muse isn’t just some generic agent, it’s sitting on a treasure trove of data and meta can build a trillion dollar business on top of it.

i like this a lot. partnering with big public companies like accenture, deloitte etc to do 3rd-party evals of misaligned AI will force these teams to be more transparent with their findings (easier to hide in a small independent startup)

Openai is now offering advertisers a feature that uses ai to re-write their ad to match the language, tone and style used by the target user in their conversation. they're also rolling out 'sponsored agents' that activate when you click on an ad. ask it any question about the product and it's goal will be to answer and convert you. its hard to see this not becoming the dominant way advertising happens online. incredibly personal, real-time ads that adopt to the way you're thinking and talking - built on millions of rich context in chatgpt.

fascinating: anthropic is building an automated biological research lab that will be steered by claude and humans - it will not create designer drugs but focus on higher advances in biology (medical, manufacturing etc) as part of this they're investing huge amounts teaching claude how to operate robots. the goal would be an end-to-end wet lab that makes frontier breakthroughs in the field. the lab is in early stages and isn't running any clinical trials (yet). so now both dario and demis are leaning heavily into ai-led medical research. demis has shifted focus to google's designer drug lab isometric labs.