No one changes the world alone Introducing Ampersand - the first model built for us. Ampersand is a collaborator trained to help people achieve their individual and collective goals. In multi-month simulations, users collaborating with Ampersand accomplished more than with Fable and other frontier models Available to everyone through our API

Florida’s ambition abounds. Enjoyed speaking with the @AmbitionAccel cohort, backed by Related Ross and the @FloridaC100. Thanks, @abramschott, Jason Conley and Greg Maiatico, for a great conversation.

For AI to work with us, it needs to understand us Today, we're introducing Persimmon, the first large-scale model designed to realistically simulate how people talk and interact

The Inevitability of the Company AI Harness At some point in the not too distant future (probably less than 6 months), the CEO of every business will be faced with answering this question: would you like to one-click install superintelligence into your business? When we reach this point, adopting AI will no longer be a process of asking individuals in the business to use Claude Cowork or ChatGPT Work to help them be more efficient in their work. Instead, it will be a decision that management teams and boards make for the entire organization. The harness will be installed at the company level, and the harness will enable AI models to drive their own adoption and deployment into the business. With access authorized at the highest levels, the company harness will: 1) Connect to all of the systems of record that contain the artifacts and assets generated by the operations of the business: email and chat, document storage, source code repositories, accounting systems, CRM, marketing campaigns, purchase and revenue data. 2) Ingest and analyze all of the external-facing activity that matters to the success of the business in its market. 3) Use the data it finds in these systems to identify and synthesize the skills needed to replicate the external-facing activities of the business to human-level quality or better. 4) Correctly assess that the vast majority of internal-facing activity in large companies is make-work, and should not be replicated. 5) Find the narrow set of processes that do actually drive the critical external outcomes, and seek to replicate and improve those. 6) Treat the creation of the automated skills and workflows that matter as a reinforcement learning learning problem. 7) Use historical data as the reference set and build itself a loop that trains a skill (or perhaps eventually a bespoke model) to deliver superhuman performance in that task. You might say that AI will never be able to learn how to do all the things a company does without humans hand-crafting the skills or teaching it their proprietary processes. That take will not age well. You might feel that this is a dark future. I don’t agree, but it won’t matter. The incentives to adopt the company harness will be too strong to resist. You might worry that this will be massively disruptive to the way that most companies operate, and you will be right. When company harnesses land, the nature and pace of AI adoption will be a difference in kind, not merely degree. Up until this point, AI diffusion has been limited by a natural brake: the rate that humans in an organization can absorb new tools, translate their workflows into agent skills, and supervise/direct the operations of their agents. The introduction of the company harness will remove that brake. It will be designed for a world where AI is doing the work for us instead of doing the work with us. The company harness will bypass the rate limits of organizational diffusion in favor of a recursive self-improvement loop that is bottlenecked only by computational constraints: tokens, data, and eventually, energy. It will be simultaneously be both thrilling and terrifying, depending on how you position yourself. Some will seek to accelerate it, others will seek to slow it down, but either way adoption will be inevitable.

This might be the most accurate @BloombergBeta review I’ve ever seen. Thanks, @einarvollset ! 😂

“Maybe someone in this room becomes the first female president.” Gloria Steinem said this while looking around her living room during NYC Tech Week. A good reminder that while technology is an enabler, people are the catalyst. The future will be built by both.

In Gloria Steinem's living room again this week with @FortuneMPW, talking AI and leadership. One idea: this is a moment women are built for. Leading people and automation takes muscles many of us have used our whole careers: juggling, empathy, knowing what only humans do well.

Big congrats to @Bluefish_AI 🐟 who announced a $43M Series B led by @thresholdvc & @NEA with @BloombergBeta continuing from the start. Attention is shifting. Bluefish makes it possible for advertisers to adapt. In <2 years, 10% of Fortune 500 are customers. Smart not to wait.

vers.sh is now open beta! we got tired of everyone being proud of running 5 claude codes at once. we think you should be running 500.

When I talk to researchers at frontier labs and AI startups, our conversations almost always converge on the same bottleneck: data. Specifically: * finding real world datasets that reflect the tasks we want AI to perform * designing evals and benchmarks that accurately measure capabilities we care about * understanding which data actually moves the scaling curve Love seeing more focus on the data layer of the stack; excited to see what @withprotegeai builds with DataLabs!

New 2026 mission: Hang out in 50 startup offices by year-end in SF/Bay Area. Friends.. who’s got an open door? 😇 I don't use Twitter much, so please tag founders or teams with cool spaces!

Today, we are thrilled to announce that @withprotegeai has raised a new $30M funding round led by @a16z. When @BobbySamuels and former @DatavantHQ co-founder and CEO Travis May first started Protege in 2024, we saw three bottlenecks to AI’s progress — compute, models, and data. Compute has scaled dramatically, models continue to improve, but access to the right data remains the hardest part. That’s why Protege exists: to be the infrastructure for real-world data in AI development. In the few months since our last round of funding, we’ve felt the urgent demand pull across industries, domains, and modalities. We are rapidly expanding to meet those needs. When AI builders come to Protege, they’re looking for real-world data: the most authentic signal of how people and systems actually behave. This is not synthetic data created by AI nor manufactured data created to simulate human behavior. Across every stage of the AI development lifecycle — from pre-training to post-training to fine-tuning to evaluation — AI builders need this data. They’re looking across modalities and industries: healthcare, video, audio, motion capture, gaming, manufacturing, life sciences, real estate, finance, education, and many more. Foundational, multi-modal model-builders (including the majority of the Magnificent Seven) now work with us across multiple domains along with dozens of other model builders. Thanks to @daisydwolf and the rest of the a16z team for joining us on this journey as well as our existing investors also participating in this round — @nbt at @footworkvc, @saarsaar at @CRV, @BloombergBeta, @flexcapital, Shaper Capital, and others. Lastly, a huge thank you to the @withprotegeai team for making this all possible. More details in 🧵! #AIDevelopment #DataInfrastructure #a16z #StartupFunding #Protege

Was fun seeing an elite athlete get into vibe coding because you could see the obsession build up and although we were supposed to shoot for 15 minutes he spent hours building things. Just generating ideas and trying them and he quickly got a feeling for what works.

Can @SHAQ vibe code an app with no coding experience? Vibe coding is easy to start with, but like any other skill, you can learn to master it. That’s why we partnered with @tadeoyerinde and the @campusgrad team to build Vibe Coding 101 for folks who want to master Replit.

Founder & CEO of @Replit @amasad on the 1000x engineer: “The buzzword in Silicon Valley has been the 10x engineer. 2-3 years ago, I talked about the 1000x engineer, highly leveraged by autonomous agents." "You're not just one person, you’re a team of engineers. Every engineer is a sort of manager now." "We're not at 1000x yet. But pretty soon, the difference between a great engineer who knows how to use AI tools and an okay or good one is going to be 100x.”

Come join us at Startup Festivus! 12/5 in SF