Partnering with Preview: Lights, Inference, Action

There’s something magical about watching a developer in flow state, fingers flying as they type into Codex, Claude, or Cursor. A powerful creation interface unfolds in response to every keystroke, powering a workflow that would have been unrecognizable, even alien-like, to any professional software engineer just two years ago. The entire IDE has been torn down and rebuilt from scratch for the age of AI, and developers are flying as a result.

Video models have raced ahead over the same two years. Yet unlike coding, video creation remains stuck in the stone age, with its entire toolchain still built around capturing, uploading, storing, and editing enormous files.

Generative video is at a threshold moment: the models are now expressive enough to become the new camera. The bottleneck is no longer the technical art of animation or the $100M budgets for film crews and visual effects; it’s human creativity and token inference. But video creators are not yet flying.

Without a proper AI-native tool suite, or a “Cursor for video,” the people who make films, TV series, and commercials for a living remain stuck bolting model outputs onto legacy editing pipelines. Generation, review, and production tracking each live in a separate application; workflows buckle under the sheer volume of material; and a finished take often can’t be traced back to its source for legal clearance. It’s a gaping problem and is the biggest roadblock on the industry’s path to truly embracing AI.

Preview clears that block. An AI-native video creation and production platform built for professional creators, Preview combines a video timeline with an infinite canvas for ideation into one collaborative multiplayer workspace. Teams can generate with any model and track every character, location, and prop, with complete records of how every asset was made. It enables both halves of a hybrid production to work together seamlessly. There is simply nothing like it in the market.

Co-founders Stefan Fejes and Veljko Tornjanski have been best friends since high school in Serbia, and were seemingly born for this mission. They were founding team members at video creation pioneer Veed; later, at Vizcom, Veljko was the first hire and became Head of AI, while Stefan led Product. They understand well the tricky nuances of video, they have deep empathy for their customers, and they have delivered sophisticated, polished work on impressively short timelines.

More than 100 studios are already actively using Preview, from agencies making commercials for Fortune 500 brands to AI-native studios to Hollywood feature filmmakers. Simply put, they love it. Customers tell us it has become the most significant tool in their work, and the single source of truth from which all downstream inference flows. 

We are at the precipice of a video token tsunami, and many billions of dollars of inference will run through this category, driving the next wave of iconic films and IP. Value will accrue to where the work actually happens—the platform a creative team opens in the morning and closes at night, where the decisions get made and the record gets kept. We believe Preview is that platform, and that this is the right team, the right product, and the right moment. We are thrilled to partner with Stefan and Veljko and to lead their seed round.

Partnering with Corma: Closing the Defensive Cybersecurity Gap

Cybersecurity is a perpetual fight between the good guys and the bad. The age of agentic AI and Mythos-class models has tipped the scales in favor of the bad guys, opening a Pandora’s box for offensive capabilities. An explosion in CVEs has since ensued. 

Cyberattacks are about technical exploitation, which requires coding and reasoning prowess. Anthropic’s Mythos can discover and exploit zero-day vulnerabilities. Any model paired with a harness brings the cost of social engineering attacks down to near zero, unleashing them at unprecedented scale. The danger is heightened with every new model release. 

Defensive security isn’t improving in parallel. The relevant data lives in logs, events, traces and telemetry, a fundamentally different modality than the bodies of text LLMs usually train on, which barely show up in pretraining. Unlike offensive cybersecurity, which involves goal-directed reasoning towards a defined objective, defense requires open-ended reasoning, continually finding anomalies across vast amounts of normal-looking activity, a distinct capability that typically demands extensive domain-specific training. In short, defense is out of distribution for general-purpose frontier models: neither the data nor the intuition it requires is well-represented in training. In the race toward AGI, we cannot expect the frontier labs to take the detour of overhauling their training pipelines for defensive cybersecurity. And we certainly can’t afford to bet our security on a hope that they do.

Corma has illustrated this “defense gap” in testing. They ran red/blue team simulations where an attacker plants a hidden backdoor and a defender tries to find it. The defender failed to find the backdoor 78% of the time, even when the defender was an identical copy of the very same model that planted it. Holding model quality constant, attackers will always have an advantage.

The only answer to scaling laws on offense is better scaling laws on defense. Recent events have shown that frontier AI for defensive cybersecurity is one of the crucial unsolved problems of the AI age. Solving it is the generational mission Corma was founded on.

Corma founder and CEO Alon Pluda and his team are training a foundation model to power defensive cybersecurity agents. Cybersecurity, like Go or chess, is a two-player, zero-sum game with a clean reward (were you breached or not?) and an endless supply of games to play. Reinforcement learning and self-play have produced superhuman results before in this kind of paradigm. We are seeing history repeat itself. Corma pushes this further with large-scale reinforcement learning across cybersecurity environments that replicate real enterprise networks, with all their tools, telemetry and noise – as part of a training pipeline that produces frontier defensive capabilities, outperforming general-purpose foundation models, with much lower per-token inference costs. 

In the cat-and-mouse game of cybersecurity, we believe Corma’s vertical integration and “sovereign AI” approach will win. Cost matters (always-on inference costs can rack up quickly). Owning the model weights matters (they aren’t subject to restrictions on cybersecurity-related usage set by the closed model labs). And specialization and inference speed really matter (fastest to protect against a new threat wins). 

Corma’s foundation model is deployed and productized as an agentic Security Workforce. These workers operate across different security tools and take on roles spanning the entire security organization: from security operations and identity management to cloud and network security and beyond. They are on the job at Fortune 500 companies and large enterprises across healthcare, finance, critical infrastructure, retail and more.

Corma’s foundation model powers agents that complete defensive work end to end. We spoke to one CISO whose Corma agent notified him of a pending attack through his Garmin watch while he was on a walk with his dog in the evening; with one confirmation, the agent was able to shut the attacker down, all while he was still walking. In another instance, within its first hour on the job, Corma uncovered, contained and remediated an active attacker campaign inside a customer’s network that their security team had missed for 52 days. 

Training frontier foundation models for defensive cybersecurity requires world-class hackers and world-class researchers – almost no company on earth has both. Alon is personally one of the most elite hackers in the world, and has assembled a cohesive and interdisciplinary team of hackers and researchers that is hill-climbing on technical challenges and building impressive commercial momentum. 

We are thrilled to lead Corma’s seed round and to be their partner as they chart new territory in what is sure to be one of the most dynamic races in AI.

Cyera and Oasis: Stronger Together

In 2021, before Oasis Security was even founded, I sat down with co-founder Danny Brickman and came away thinking he had something familiar. He showed the same drive we’d seen in the founders of two other Sequoia companies, Wiz and Cyera, whose founders, like Danny, had come up through Talpiot, the leadership program of the IDF. I wrote about that meeting a couple of years later in my investment memo, when we backed Oasis Security, at the Series A. I didn’t realize at the time how literally connected that comparison would turn out to be. Today, Oasis is joining Cyera — and it turns out the founders’ stories make the case for this combination almost as well as the technology does.

Three years ago, the pitch for Oasis was simple: enterprises were about to have far more machine identities than human ones, and almost nobody had a real way to manage or secure them. That was true then. It’s far more true now. Every AI agent creates or uses non-human identities — API keys, service accounts, OAuth tokens, agent-to-agent credentials — that act with real permissions and real blast radius, and that don’t behave anything like the human identities the last generation of identity tooling was built around. AI drove non-human identity from a niche IAM subcategory to a board-level CISO conversation. Danny Brickman and his co-founder Amit Zimerman built Oasis, which they call agentic access management, for exactly that world.

Cyera arrived at the same shift from a different direction. Yotam Segev, Tamar Bar-Ilan, and Yonatan Itai built the company to answer a deceptively simple question: where does an organization’s sensitive data actually live, who can reach it, and what’s really at risk? As AI agents started reading, writing and acting on that data, they reached a conclusion that now looks obvious: securing AI means securing the data, the identities acting on it (human and non-human), and the agents themselves.

Put the two products side by side, and they complement each other perfectly. Cyera knows where the sensitive data is. Oasis knows who or what can actually touch it — every service account, API key and agent credential, and whether it’s behaving the way it should. It’s a bit like a bank vault: the best way to protect it is by knowing both what’s inside and who’s approaching. Data security tells you what’s actually at stake. Identity security tells you whether the thing touching your data is legitimate. Together, the two products cover the entire path an AI agent takes through a company’s environment, from the data it touches to the identity it uses to touch it.

There’s a practical case here too. Cyera has built one of the most powerful enterprise GTM teams, with the kind of scale that usually takes a startup a decade to reach. Oasis’s product now gets to run through that engine, into enterprise accounts it would otherwise have taken years to earn on its own. And the market isn’t waiting: Whoever builds the first genuinely complete AI security platform, spanning data, identity and agents, has a real shot at becoming the default layer enterprises standardize on as agentic AI rolls out. Speed and completeness both matter, and the combination of Cyera and Oasis creates both.

The founders’ backgrounds make the fit feel almost inevitable in hindsight. It’s a small world: two teams we’d backed independently, drawn from the same pool of Israeli technical talent, ended up building the two halves of the same answer to AI security. Talpiot and Unit 8200 alumni have produced a disproportionate share of the last decade’s big cybersecurity companies, so the coincidence is smaller than it looks, but it’s still a good sign when it plays out this cleanly.

Danny, Amit and the Oasis team built the category-defining company in non-human identity in under three years, from a standing start to a product enterprises trust with agentic access at scale. That doesn’t happen without genuinely hard technical work, and a founder team that saw the shift to agentic AI before almost anyone else in security did. We at Sequoia have backed both Cyera and Oasis since their Series A rounds, and we’re proud to keep backing this team as they take on the next chapter together — a bigger stage, and an opportunity sized to match.

America’s Open-Model Paradox

To Distill, or Not to Distill?

China increasingly supplies the models Western companies use to serve, train, and build AI.

Qwen’s share of new open-model fine-tunes and adaptations rose from 1% in January 2024 to 69% by February 2026 according to ATOM’s Report. The majority of American AI startups seem to be using Chinese open weights somewhere in their stack.

This dependence now extends upstream. Western application companies are building on Chinese Open Source models. Further, Western labs are using Chinese models as teachers and sources of synthetic training data in the torrid race to close the frontier gap.

For example, Thinking Machines pre-trained Inkling independently, but used synthetic data generated by Moonshot’s Kimi K2.5 to bootstrap its supervised fine-tuning.

The relevant point is not how much of Inkling came from Kimi. The point is that a Western lab had a legal path to learn from a Chinese open model, while equivalent use of GPT or Claude outputs is prohibited.

The flow today looks something like this:

• Western frontier models → alleged unauthorized foreign extraction → Chinese open weights → lawful Western post-training

The missing direct route is:

• American frontier models → lawful Western post-training

Why does that matter? Pre-training creates a capable base model. Post-training turns it into a useful coding, reasoning, tool-using, and agentic system. A stronger teacher converts part of that expensive discovery process into a cheaper learning problem.

Distillation does not explain China’s entire open-model lead. Chinese labs have world-class researchers, substantial compute, strong pre-trained models, software-hardware codesign, and rapidly improving post-training capabilities. But distillation compresses the costly final gap between a strong base and a near-frontier system. Even if distillation represents a smaller share of  a Chinese model’s total capability, it represents a meaningful share of its advantage over American open models.

New enforcement mechanisms will make large-scale distillation harder, slower, and more expensive for Chinese companies. However, enforcement will not eliminate distillation baked by state actors. Every Western frontier advance therefore creates another teacher for Chinese labs. Western builders must either reproduce those capabilities independently or wait to learn from Chinese models.

This gap gives Chinese labs a recurring structural advantage over Western companies.

The stakes extend far beyond model revenue. Suppliers of the open layer become the default base for products, synthetic data, post-training systems, evals, agents, optimization, and applied AI.

The prize is to become the substrate on which global enterprises build and improve digital intelligence.

The Weights Are Open. Our Dependency Is Not.

Downloading a Chinese model gives a Western company control over the particular version. It can run the model locally, modify it, and continue using it without permission.

But AI capability is an upgrade cycle. Western startups, model developers, and researchers increasingly rely on each new Qwen, Kimi, GLM, or DeepSeek as a stronger base, a teacher, a source of synthetic data, and a platform for further research.

If China stops releasing its strongest models, existing products will not break. They will fall behind. Reuters recently reported that Chinese authorities have discussed restricting overseas access to advanced models, including models that have not yet been released. No final policy has been announced, but Western access ultimately depends on Chinese labs and regulators continuing to publish.

There is also a more technical security problem. An open-weight model is not necessarily an auditable model.

The weights are the compressed result of training. They do not reveal the full pre-training corpus, which data was filtered or poisoned, what interventions were made during training, or whether rare trigger-dependent behavior was embedded.

We are not alleging that Qwen, Kimi, or another Chinese model contains a backdoor. The point is that possessing the weights cannot prove the absence of one. A backdoor can remain dormant during ordinary testing and activate only when an unknown trigger appears. Research shows that deliberately implanted behavior can survive supervised fine-tuning, reinforcement learning, and adversarial training.

That may be an acceptable supply-chain risk for many consumer applications. It is not acceptable for defense, intelligence, or critical infrastructure.

Open weights provide control over deployment. They do not guarantee continued access to better models, nor alignment, trust and safety in the model itself.

A Framework For A Direct American Path

American companies need a legal way to turn American frontier capability into cheaper, ownable models.

Without a domestic route, the West may lead at the closed frontier while falling into dependence on China for the open layer.

A useful framework has three parts.

1. Keep building Western base models: Reflection (building open super-intelligence for enterprises & sovereigns), TML, and Nemotron are making incredible progress. But stronger pre-training alone does not solve the teacher problem. American builders also need a lawful way to absorb capabilities already developed at the American frontier.
2. Create controlled teacher access: Frontier labs could sell structured training rights to qualifying Western and allied companies, whether the resulting models are released openly or deployed privately. Access could trail the frontier, cover defined capabilities, be limited to verified companies, and be metered and audited. The most sensitive biological and cyber capabilities could remain restricted. This would not allow companies to clone the newest frontier model. It would create a legal, priced route for capability transfer that sophisticated foreign actors are already pursuing covertly.
3. Keep raising the cost of foreign distillation: Better identity verification, access controls, proxy disruption, and enforcement should continue. If a voluntary market does not develop, access could eventually become a condition attached to major federal AI contracts, for example.

These are starting points. Who qualifies, how far access should trail the frontier, how it should be priced, and which capabilities remain restricted are topics that deserve real debate.

Imagine if we had not allowed for training on the open web. We would have no leading AI at all. These types of policy implications are transformative. All the leading labs benefited from copious amounts of openly available data. We have to have an open and free future: It is imperative for Western competitiveness.

What should no longer go unquestioned is the current equilibrium: the West creates the frontier, part of that capability travels indirectly through Chinese models, and Western builders then depend on those models to make intelligence cheaper, adaptable, and sovereign.

To distill or not to distill is not the question. The question is whether the West creates a legal domestic path for capability transfer – or  relies  on an indirect path through China.

Partnering with Etched: Building the Inference Machine

Inference is on the path to becoming the largest market in the world.

If this AI dream is everything we hope it to be – and so far, the signs point to ever-increasing acceleration of capabilities – then inference will power every storefront, every movie, every medical encounter, every video game, every lawsuit, every line of code.

Today, the inference market is so obviously large that building an inference-specific system is consensus. Back in 2022, when Gavin, Chris, and Rob were still in their dorm rooms at Harvard, that bet was deeply contrarian. The result: Etched is the only post-ChatGPT-era hardware startup with production-ready custom silicon, ready to ship in 2026. 

Over the past years, Etched has pioneered several research breakthroughs – low voltage inference and cluster-scale memory – each a hard-won architectural choice that drives step-change gains in throughput and latency. As the unit of compute has moved from the chip to the rack to the cluster, Etched has designed for that reality, building not just chips, but cluster-scale inference systems. Driven by these fundamental innovations, Etched’s inference system excels in throughput and interactivity on the full span of frontier models – from large sparse MoEs, to dense transformers, to alternative architectures entirely, like Mamba. 

Believing this is one thing. Building it is another entirely. Novel computing hardware is among the hardest things in the world to get right, and the ground underneath is constantly moving. Models change every few weeks, context lengths stretch, attention gets reinvented, the mix of dense and sparse keeps shifting. Winning here means doing two things simultaneously: iterating fast enough to keep pace with the models, and pushing the frontier of what the hardware can do. That takes a rare kind of team: visionary enough to create the right designs, young enough to move fast, proven enough to have chips in production, and built to do this for decades, not a single tape-out.

Walking into the Etched office for the first time is like a bolt of lightning. Go in the morning or at 11 p.m. – the energy is the same, and it is infectious. Gavin and Rob are the kind of outliers both daring enough to design silicon from first principles and bold enough to knock down every obstacle in their path, and the team they have assembled matches them. We have watched them do the things that separate those who merely talk about hardware from those who ship it: standing up a live lab in San Jose to host their first racks; opening an office in Taiwan to colocate with suppliers and expedite testing; questioning every single assumption for what is possible, such as low voltage inference. “Production is the product” is the company’s mantra. Shipping is all that matters. 

The company’s progress has been rapid. Earlier this year, Etched taped out its first generation chip at TSMC, making it the first post-ChatGPT-era company with a successful full-reticle A0 chip tape-out on TSMC’s leading-edge nodes. Then, in a span of just 40 days, the Etched team brought up their first cluster of chips to run inference on a wide range of frontier AI models, achieving Pareto dominant performance on industry-standard throughput-interactivity curves. Now, Etched’s earliest customers are getting access to the product, seeing for themselves the speed, throughput, and expressiveness of the system. 

Etched has built a beautiful machine in Gen 1. We expect it will do very well in the market. But we are partnering with Etched because we believe they have built the rarest thing in this industry: the machine that builds the machine. This is a team that has the taste and the relentless execution to keep shipping the next generation of inference machines, faster and more ambitious each time.

The intelligence race is fundamentally compute constrained. Pushing the limits of physics to deliver the maximum intelligence per watt is both an insanely fun engineering and operations problem, and an incredibly noble mission for humanity.

We are delighted to be partnering with Etched and leading their $300 Million Series C at a $10 Billion pre-money valuation, joined by our friends at Jane Street, Andreessen Horowitz, Diffusion, and SK Hynix.

Partnering with Bunkerhill Health: AI Agents that Improve Patient Outcomes

Healthcare hasn’t historically been known as a fast or early technology adopter. Researchers struggle to access external data, collaborate with research institutions, and navigate complex regulations. Hospitals and health systems try to create their own tools or implement narrow point solutions from vendors, but they too often stall in experimentation and never reach patient care. 

Health systems need the right platform to quickly adopt AI, so they can refocus their time back on improving patient outcomes and providing better care.

Bunkerhill Health is that platform. Their core product, Carebricks, lets health systems create and deploy AI agents across any clinical or operational use case – turning the data a hospital already generates into timely action for the patients who need it. We first backed Nish Khandwala and David Eng at the seed, and we are proud to continue to double down. 

Nish did not read about this problem; he lived it. As a graduate student at Stanford, he helped build an algorithm that could estimate a patient’s coronary artery calcium score from a routine chest scan – a signal of cardiovascular risk that normally requires a dedicated, more invasive test. The algorithm was good enough to be published in Nature Digital Medicine. And then it went nowhere. Around the same time, Nish’s father suffered a heart attack, and his CT scan carried the very calcium signals the Stanford algorithm was built to catch. It is one thing to know that proven science sits unused in academic papers. It is another to watch it fail to reach your own family.

That experience taught Nish something non-obvious: the bottleneck in healthcare AI is not the technology, it is everything around it. So Bunkerhill initially built a system that gets an idea from research to the bedside – sourcing data from a consortium of leading academic medical centers, validating the model, shepherding it through FDA clearance, and installing it into the hospital where a clinician and a patient can actually use it. But they ran into another problem: health systems do not have the time or resources to deploy one-off AI solutions. So Bunkerhill built Carebricks to allow any health system to bring their ideas to life with AI agents across clinical, operational, and administrative work in one central platform. Health systems bring the ideas and the clinical judgment, and Bunkerhill turns them into agents.

We have watched this compound in the field. At UTMB Health, Bunkerhill rapidly went from one agent in production to more than twenty, spreading from one specialty to the next as clinicians saw what it could do. And just as importantly, it has allowed UTMB to consolidate what used to be multiple vendors’ point solutions into a single platform with Carebricks. 

Bunkerhill’s agents run quietly in the background, surfacing the findings that matter and catching the cases the system would have missed. Not to automate clinicians away, but to automate manually-intensive work so clinicians and their teams can focus more time on what they do best: caring for their patients. 

Nish and David are outlier founders with the technical depth and the relentlessness this problem demands, and they are pulling healthcare toward faster iteration and better outcomes for every patient, not just the average one. We are excited to partner with Nish, David, and the entire Bunkerhill team as they build the agentic AI platform for healthcare.

Partnering with Sable: Closing the Diffusion Gap

The frontier of AI is moving at a pace most of the economy can’t match. Labs ship new capabilities every month, while the Fortune 500 scrambles to absorb what came out last year. Watching this gap widen, it’s clear that the bottleneck isn’t only intelligence, it’s helping customers understand what AI can actually do for them. It’s the reason we’re thrilled to partner with Sable.

Imagine your business wanted every buyer to fully understand your product or service, without the realities of time and scale. What would you do? You’d put your best product expert in every conversation. They’d learn each customer’s goals and deeply empathize. They’d demonstrate the product instead of describing it. They’d answer every question, adapt to every level of experience, and stay until the customer understood not just what your product is, but what it could do for them. And they wouldn’t do this for your largest accounts alone. They’d do it for every prospect.

Sable makes this possible. It created “Aidan,” the first AI employee, who leads its own customer calls with vision, voice, video and real-time browser interaction. Sable is less than one year old, but frontier companies are already using it to explain their products, including Notion’s custom agents and Decagon’s agent-building platform, alongside large public enterprises. There is a staggering range of applications and organizations eager to deploy AI employees, from the world’s fastest-growing companies looking to keep pace with demand, to large enterprises seeking to accelerate growth with AI. We’ve been blown away by the market pull for this new capability, with +150 companies already on Sable’s waitlist.

Enterprise deployments of multimodal AI employees require a combination of customer obsession and technological breakthroughs at the frontier of the field: low-latency browser use, real-time vision and simultaneous human-agent interaction. Problems like these are ultimately won by people. Sable is exceptionally well-positioned to solve them.

The four founding team members are friends from Harvard, where they studied post-training, reinforcement learning and multimodality. They worked at SpaceX, Google, Meta and Together AI. Nim, co-founder and CEO, is a rare combination of a leader and visionary. He has an innate talent, and impressive experience recruiting the highest-caliber people to work together and achieve the unlikely. That philosophy has shaped Sable from the beginning: the company is obsessed with hiring exceptional talent. Among their first hires is a unique mix of applied AI researchers and customer-obsessed engineers, including ten Harvard alums, former quantitative traders and International Math Olympiad winners.

The combination of an all-star team, undeniable market demand and bold vision is why we were thrilled to lead Sable’s seed round and co-lead its Series A. We believe Nim, Leon, Linda and Itamar are building foundational infrastructure for companies to manage their AI employees, and we couldn’t be more excited to be on this journey with them.

Partnering with Probook: AI for the Trades

There’s no substitute for authenticity. In a world of AI, real, genuine human connection matters more than ever. The story of Probook starts as real as it comes. 

After decades on the job, John, a New York City police officer, was injured on duty and couldn’t serve on the force any longer. Some time later, he took up power washing to stay busy and tapped his son, George Eliadis, to help alongside him. George worked weekends and summers in high school, then continued to power wash while he pursued the prestigious M&T dual-degree program at the University of Pennsylvania.

George noticed how much logistics got in the way of what his dad loved most: power washing. Years later, George spent time inside TR Miller, a home services business in Illinois that became Probook’s first customer, and saw the same problems at scale. He learned that what slowed his dad down affected millions of electricians, HVAC techs, plumbers and other workers. The administrative load, from assigning jobs to following up on website inquiries, cuts into already-tight margins and hurts both customer and employee satisfaction. 

AI was supposed to take all that on, but reality hasn’t lived up to the hype. Operators bought a tool for every task, and ended up with a stack that couldn’t talk to itself. Nothing connected; costs ballooned. Customer experience stayed broken, and no one’s life got any easier.

George realized his dad’s life—and an entire industry—would benefit from better, more connected software. So he and co-founders Lewis Zhang and Ben Cervantez launched Probook, the AI Operating System for home services—built around dispatch, the foundation everything else depends on. It is a single platform that shares context across the entire customer experience, automating from intake to dispatch and beyond, and expanding margins for the businesses that keep America’s homes running. Probook is personal for George, of course—but it’s not only George who takes this work personally. All three founders are the real deal. Their authenticity shines through, and it’s the reason why we at Sequoia wanted to partner with them from the seed round and now in their Series A.

Probook works like this: your water heater goes out, and you call a local plumbing company that runs on the platform. Probook’s AI picks up immediately, already knowing each technician’s experience, availability and distance from your home, along with their close rates and ticket sizes. It assigns the right tech to the job, alerts them, and keeps you in the loop with an ETA. You get fast help; the company runs leaner; its team earns more with less stress.

Since childhood, George, Lewis and Ben have been scrappy and relentlessly curious. George is a near lifelong entrepreneur—between power washing jobs, he sold everything from coins to vinyl records. He attended Penn’s M&T program along with Ben, who was valedictorian and captain of his high school football team, and interned at McKinsey during college before dropping out to join Probook. Lewis started coding when he was 7, earned degrees in Electrical Engineering and Computer Science from Berkeley in two years, and was part of the critical player-server matching team at Roblox. Dedication is what drives the team: to say they show up for customers is an understatement. They care deeply about tradespeople and are obsessed with taking the drudgery out of their work, flying all over the country to onboard and work with new businesses in person. And for those companies, Probook has been nothing short of life-changing.

Probook’s growing team is scaling quickly, with a mission to transform the trades. When the nerve center of home services is streamlined by AI that actually drives outcomes, operators can run better businesses and build better lives for their staff. We are proud to support George, Lewis, Ben and team.

Listen to the Market

I’ve always been fascinated by solving problems. First, they were math problems in primary school. Then, math, physics, and engineering problems in high school. Then, applying math to decision and control problems in college. Then, applying statistical methods to forecasting problems. Today, I’ve moved on to solving complex company-building problems, but when you break down most complex problems into simpler ones, they often come down to predicting risks and forecasting expected value. Over the years, I’ve learned that the best people, groups of people, algorithms, or systems of algorithms for forecasting the future are full-spectrum signal processors rather than narrow-band signal amplifiers.

Try a small experiment. Ask a dozen random people whether the country is drifting left or right, and you’ll get a dozen confident, incompatible answers, most of them shaped by whichever sources that person already trusts. Media outlets are competing for eyeballs. Polling tells you what people are willing to say out loud. Your feed tells you what your particular corner of the internet wants to be true. None is a reliable guide to what actually comes next. We are buried in political information and starved for political clarity.

That gap is widening. We are wired to seek out what confirms the story we already hold. Year after year, the payoff for being loud has outrun the payoff for being right, and we’ve sorted ourselves into audiences that rarely see the same facts, let alone agree on them. When it is time for political debates, we have already picked a side, and our natural inclination is to root for our team rather than get to the truth.

One mechanism has been consistently strong as a full-spectrum signal processor rather than a narrow-band signal amplifier: the market. Markets make people back beliefs with capital, and capital has a way of sharpening judgment that pundit commentary never will. Anyone can hold a view for free; holding a financial position costs you if you’re wrong. What comes out the other end is a price: a continuously updated consensus of everyone with money and conviction on the line, stripped of the performance that dominates everywhere else. If you care what is likely to happen rather than what is smart or fashionable to say, the price is the better witness.

That is the case for what Kalshi launched: the Kalshi American Power Index, or KPOW. It is an S&P 500 for politics. It compresses a vast, churning system into one number you can track. KPOW runs on a scale from +50 on the Democratic side to +50 on the Republican side, and it blends two layers: a quarter of the weight reflects the certified reality of who controls the House, Senate, and presidency today, while three quarters reflects what Kalshi’s markets imply about who controls them next, assembled from six pieces, ranging from chamber control to expected seat margins to the odds of a government shutdown. The mechanics matter less than what they produce: a single line that moves when power actually shifts, not just when people comment on it.

What we find most compelling isn’t where the number sits on a given day; it’s what happens when it moves. A poll is a still photo taken on the afternoon someone happened to ask. An index is a continuous recording. A court ruling, a primary upset, or a shock from overseas shows up as a bend in the line you can point to and size. That converts vague arguments into something testable. Did the redistricting decision actually move power, or did it just own a news cycle? You used to be able to debate that endlessly. Now there’s a measurement to debate against.

Prediction markets built their reputation on questions with a finish line: a winner, a vote, a clear settlement. The questions that shape how we feel about the country mostly don’t have a single answer. “Which way is power tilting?” never resolves; it simply keeps shifting. An index is what allows a market to address a question like that, not by crowning a winner, but by returning a number, the way an equity index distills an entire economy into a single readable figure. Kalshi could already price the questions that end. As of today, it can price the ones that don’t.

And none of this exists without the exchange underneath it. An index is only as trustworthy as the markets feeding it, which means it requires the kind of liquidity, regulation, and settlement infrastructure that takes years to build. Kalshi has spent those years. And once that foundation exists, an index is just the first instrument it supports. The same layer can carry a whole catalog of contracts tied to questions that the legacy financial system was never designed to touch.

The defining questions of the coming decade, across politics, policy, technology, and markets, will mostly be the messy, never-quite-settled kind, and they’ll arrive buried in more noise than ever. The opportunity is to give people a way to measure those questions rather than just shouting past one another about them. That’s why we’re proud to back Tarek, Luana, and the Kalshi team. The contest for power will keep swinging back and forth. For the first time, we can actually watch it move.

Explore the KPOW here.

Although Cameron McCord wasn’t himself present at Edwards Air Force Base, what was about to transpire that May 2025 day in the Mojave Desert would still make it one he would never forget. AJ Piplica, CEO of Hermeus, was huddled with his team on the tarmac. No matter the outcome of the day’s flight test, he too was going to have a company-defining day. After years of labor, Piplica and his team would be testing their new hypersonic airplane engine, first taxiing down the runway, and assuming that went well, actual liftoff. 

Even more nerve-wracking than the prospect of fundraising if the test failed was that of the dual taxi and liftoff test in a single day, a feat that would have been unthinkable even a few years ago. “Our data from the taxi needed to give us confidence that we’re not taking a dumb risk by attempting takeoff and landing. And that’s a huge amount of data to review, that historically took weeks, if not months, to parse between high-speed taxi and first flight,” Piplica says. The time they’d been allotted on the runway for both tests: two hours. 

But his team had a new secret weapon. After years of employing a messy patchwork of data review tools that were not designed for hardware testing, Piplica had recently signed a contract with McCord’s Nominal, a company whose sole focus was testing hardware for real-world deployment. Hermeus had seen success with the platform for preliminary Hardware-In-the-Loop (HITL) tests in their Atlanta warehouse, but never in the field with the Air Force breathing down their necks, and certainly never with such high stakes and so little time.

As Hermeus’s plane started taxiing smoothly down the runway just under liftoff speed, data began to stream in through Nominal’s platform: the health of the brakes, actuators, avionics, electronics, control surfaces and more—terabytes of data giving a real-time window into the system’s health. As the plane slowed its taxi on the end of the tarmac, Piplica checked in with his engineers, hoping for good news. “Data review was done by the time we’d towed the airplane back to the other end of the runway,” he remembers, “everybody was thumbs up, so we concluded that, and we were at a safe level of risk to attempt a flight.” Minutes later, their Quarterhorse Mark One was in the air for the first time. When it landed, Piplica snapped a picture and texted it to McCord, Nominal’s CEO. Under the photo it read simply, “Hey, we got it back.” 

McCord’s Mojave flight test assist was far from his first brush with the Armed Forces. He was raised on stories about his grandfather, a greatest generation archetype, who had been rushed through the Naval Academy in three years to join the war in the Pacific, witnessed the Japanese foreign minister sign the instruments of surrender, had seen nuclear weapons detonations, and flown the longest flight in history over the South Pole. McCord’s father was a federal attorney in the Department of Justice, and his mother was a special needs teacher. In their home, being of service was a mantra, and an action; most days, McCord and his siblings were explicitly urged by their parents to question how they were being a force for good in the world. For McCord, inspired by his grandfather (and uncle and cousins in the Navy), “Good,”  meant joining the Armed Forces. 

He paid his way through MIT by doing ROTC, which is where McCord “got bitten by the entrepreneurial bug,” he remembers, but more via osmosis than in practice. In addition to ROTC, he played varsity soccer, double majored in Nuclear Engineering and Physics and minored in Political Science, for practical reasons (“All of this engineering, how do I transition that into something that impacts the world?”). Between training and practice and problem sets and lectures and more lectures, McCord observed his classmates tinkering and creating. In particular, he remembers not infrequently passing by his fraternity brother, Jason Hoch at 4 a.m.. Hoch would still be awake, finishing a CS problem set or hacking away at a new idea as McCord, just up and in uniform on his way to ROTC training, urged Hoch to get some sleep. He admired his classmates’ will to build and create, knowing it was an exercise he couldn’t yet fully embrace, but one that he filed away to explore in some eventual, less time-constrained future.    

Within days of graduation, McCord was onboard the USS Helena, SSN-725, the newest officer on the submarine. Any notion that respect would be granted by virtue of his positional authority or pedigree was quickly disabused. “You just start from scratch with first principles—how do you build trust, rapport, and respect with these people where everything you were is stripped away to zero?” says McCord. He did manage to win over his crew after they observed that what he lacked in revolutions around the sun, he made up for in thoughtful leadership, attention to detail, and commitment to understanding the nuances of the underwater behemoth they called their home. “It was my duty to understand that complexity,” says McCord. “Especially for those around me that were relying on me.” Before his service was finished, he’d have a chance to prove just how well he could navigate that complexity.

“We think it’s his appendix,” one of McCord’s crewmates informed him, as a fellow sailor nearby doubled over in pain. “It seems bad.” A few years had passed since McCord was the ‘new guy,’ during which he had experienced a midnight fire, a change in presidential administrations, and hundreds of nights underwater—but this was new. “Appendicitis on a deployed submarine on a mission is not a good thing. So the time was ticking,” says McCord. The ship’s only medically trained officer was not equipped to perform emergency surgery, and since the sub was mid-clandestine mission in North Atlantic waters, reaching a hospital before rupture meant navigating the vessel through Nordic fjords. McCord was tapped to ‘drive.’ 

Assuming the Conn (naval parlance for, “control of the ship”), he checked above water conditions: temperatures hovered around zero, and a blizzard made visibility nonexistent and conditions turbulent. But despite inclement conditions, this time, McCord wasn’t at a loss. Anxious, certainly, but his years of training, and his dedication to ensuring that he could be of service to his crew, paid off. “Because of my familiarity by then with all of the complexities of the submarine, I was able to gut into how to do this,” says McCord. 

By then, McCord was accustomed to managing machinery designed during the Cold War, but that didn’t mean he relished the challenge. “We had to do a lot of very unnatural things with the submarine to get through this fjord and open that rear hatch.” When they neared the shore, a particularly burly soldier “essentially threw the sailor over to the Norwegian coast guard, who picked him up and rushed him to a hospital,” says McCord. Twelve hours later, a WhatsApp message from their new Nordic friend gave them the all clear health-wise. 

For two more years, McCord lived many of his days underwater. In his free time, eager to stay apace with the outside world, McCord Coursera’d. “I would watch pre-downloaded Andrew Ng Stanford AI classes. This was in 2015, 2016, and I would be teaching myself and actually building out early models.” His autodidactic afternoons keeping up only reinforced how far his surroundings had fallen behind. “Learning AI, and then going into the control room where you have 1970s software, old hardware—it was a crazy cognitive dissonance.” 

 As he neared his five-year mark on the sub, it began dawning on McCord that the model of service and impact that had worked so well for his grandfather and uncle, both with decades-long careers in the military, might not be the right fit for him. “The military allowed them to make a huge impact on the world over 30 or so years,” says McCord. “But in today’s world, technology seemed like the way to make that impact. I wanted to use service-aligned technology, but I was not cool with the idea of having to wait 30 years.” 

McCord had put in his time, “but then this incredible opportunity came by,” he says. He was selected to be one of the Navy’s liaisons to the House of Representatives. Between 2017 and 2019, McCord got a front-row seat to another deeply complex system. “You get to understand how a bill becomes a law,” says McCord. “But you also get to understand the personalities, the relationships, the executive branch, how budgets get passed, the back doors on The Hill—how it actually happens. I lived that viscerally for two years.”While his job description was to “sort of be a good steward of the Navy, tell war stories, build support,” he also made time for an extracurricular: “I was someone that members of Congress and staffers could go to to understand cutting-edge technology,” says McCord. “I think it’s hard for them to find, frankly, a low-threat way to do this. I developed this reputation of, ‘Hey, you can actually just go down to the Navy liaison shop and there’s this MIT guy Cameron who can just explain LLMs or cybersecurity or why technical stuff in this bill is relevant.” In 2020, capping off his career in public service, McCord’s technical acumen earned him an invite to help develop the House Armed Services Committee’s report on how technology was going to change the nature of warfare (and world). “It’s a little bit crazy to say now, but it was really some of the first governmental writing about AI,” says McCord, whose name is listed in the footnotes among four-star admirals and heads of agencies.

From his years on the sub, McCord was no stranger to waking up at the crack of dawn, but now he found himself doing so in a decidedly more terrestrial context. It was 2020, and he and his first private sector team, Anduril’s drone defense system engineers, were cruising through California’s Central Valley towards the desert to test their system. When they found a suitably remote, dusty patch, they unpacked their hardware—a tower covered with cameras and infrared sensors and miniature drones—and set up their Wi-Fi ‘pucks’ to run flight tests. These sessions would often last days, with team members camping in their trucks to avoid the sunrise commute. Each morning, they would begin tests anew, generating telemetry and sensor data and logs and video and images. “And it was so difficult with the tools we had to capture that data, intuitively organize it, and just answer the question: Did the test work?” As the Product Manager of the system, he led his team in long whiteboard computational sessions, hours in MATLAB, or dated academic graphing software. “None of this was production scale. And none of this was modern in any sense,” says McCord. 

The process itself he enjoyed—the rolling up of sleeves, the camaraderie—all of that was pleasantly, arduously familiar. But day after dusty day, McCord remembers feeling renewed shock at the state of hardware testing. Here he was, working in a startup ecosystem where software innovation thrived on cycles of rapid iteration, but hardware testing was stalled in a different century. And it wasn’t lost on him that if Anduril, a company with “all of the venture dollars in the world and incredibly smart people had challenges getting this right, what is like at old organization X or massive company Y? In the back of my head this was just so clearly an area where better software could improve quality of life, and improve outcomes,” says McCord. “To be clear, I didn’t have the answers, but I just was obsessed with the problem.” 

McCord first attempted to solve that problem with emerging data technologies: tools built for business intelligence, data marketing, SQL-based tools. In short, nothing “built for the types of telemetry and high-frequency sensor data that robotic hardware systems generate,” says McCord. “So they don’t work.” After fifteen months at Anduril, McCord couldn’t ignore his obsession any longer. Eager to get his bearings in a new complex system—the world of entrepreneurship and VC funding—McCord pitched his rough idea, improving hardware testing, and himself as a part-time entrepreneur-in-residence, to Josh Wolf at the VC firm, Lux Capital. In return for an opportunity to speak with dozens of hardware CTOs to pressure test his thesis, McCord offered to help Lux develop a dual-use, government-private sector business strategy. 

For a year, co-enrolled at Harvard for his MBA, McCord effectively moonlit for Lux. “This was only possible because COVID was happening,” says McCord. “I would do Zoom classes and I would shut my Harvard Business School laptop and open my Lux laptop and basically alternate between the two.” HBS is where McCord first encountered Bryce Strauss, a kindred spirit with an aerospace background. McCord asked him to coffee, expecting a 30-minute conversation, and the pair ended up venting for nearly three hours. “It was this winding back-and-forth where we both obsessed about post-test analysis and data review, and how it’s essential to quickly get insight into performance data —whether you’re building an airplane, a car, a nuclear reactor, a drone, or a robot. Every person that builds hardware does this thing,” says McCord. “And we’re like, if we could just make that process 10 times better, we think we can build something valuable here.” 

Strauss concluded they needed to turn this idea into a company together, and he wouldn’t take no for an answer. “I’m always doing a lot of things, and Bryce was this incredible unifying north star that was like, ‘Cameron, when we graduate, we’re doing this,’” says McCord. The duo decided to enlist a third, software co-founder. For McCord, there was really only one choice: “I was like, ‘Hey, I know this guy, Jason Hoch from MIT. He’s the smartest software engineer I’ve ever met. I think he’s the person to be the third leg of the stool.’” Strauss, the aerospace expert, came up with a name, a play on “All systems nominal,” common parlance for “all good” during rocket launches. 30 days before graduation, Lux Capital wrote Nominal their first check. 

“There’s a bug,” Hoch said from the backseat of their rental car. Six months into their tenure as co-founders, the trio had flown to Los Angeles to demo an early prototype of Nominal’s hardware-in-the-loop (HITL) testing system for a major satellite company, who they hoped would be their first paying customer. “Something’s off with time synchronization of different data sources,” added Hoch. With 20 minutes to go before their pitch, he’d have to do some en route hacking.

When they arrived, the trio walked into a boardroom to find roughly a dozen company leaders waiting. Strauss demoed their product, demonstrating the speed with which it would allow engineers to manage and interpret satellite test data. The last minute hack held, and the time sync across different data sources—telemetry, engine health, computation speeds and more—worked. The trio was ecstatic, even when the satellite company only signed on for an unpaid pilot of their software. “They took a bet on us and they let us learn with them, frankly, which is more valuable than anything,” says McCord.

With some assurance their product worked and had utility for customers, McCord, Hoch, and Strauss decided to hire four more full-time engineers. Still the latter days of COVID in fall 2021, everyone worked remotely half of the time, and would converge in Austin every other week, rent an Airbnb, and build from wakeup to well into Wendy’s-fueled late nights. As the company grew, their Austin reunions became every third week, then once a month. “Finally, we reached a point when we had around 40 people where we declared, we’re not going to have regularly scheduled Austin weeks anymore.” 

When he wasn’t sleep deprived in Austin, McCord lived in Washington, D.C., a home that allowed him to leverage his public sector connections while growing his private sector customer base. That’s where he first met AJ Piplica for coffee at Commonwealth Joe’s. Before witnessing the pain of hardware testing as CEO of Hermeus, Piplica had experienced it in the public sector working for NASA. “Within the Department of Defense, which is a very vast test infrastructure—all the data is siloed,” he says. “You were literally moving CSVs around by hand and opening things in Excel.” After witnessing Nominal’s utility during his first Mojave flight test, Piplica recognized the step change it could represent for all hardware development. “Any organization that’s taking data out of the real world, which is, like, every major company in the world could benefit from this,” says Piplica. “Yeah, robotics and AI are cool, but what’s actually cool is when you put them together. That nexus between the digital and the physical world is what really unlocks a huge amount of growth for humanity.”

Ten days before his wedding in January 2025, McCord got a call from Alfred Lin, partner at Sequoia. Nominal was eyeing another round of fundraising to support the expansion of their team and their product. Lin, who had met McCord during Nominal’s Series A process but ultimately deferred investing (“We wanted more evidence in support of his hypothesis before investing”), understood the tailwinds accelerating Nominal’s growth, and wasn’t about to let another round pass him by. “We are living through a hardware renaissance, and we were looking for a new platform that supports modern hardware engineers on this journey”, says Sequoia partner, Anas Biad.

McCord wanted to work with Sequoia, but he wanted to get married first even more. “I told Alfred, look, I’m getting married. But can we schedule time for me to come to SF right when I get back? I will walk you through everything in the business.” Lin agreed, and as promised, McCord flew straight from honeymooning in New Zealand to meet with Lin and Biad. The partners were impressed by McCord, but told him they needed to do their due diligence on the product before any decisions were made. “For days, Anas basically didn’t sleep. He called every single one of their customers,” says Lin. In the end, McCord was reassured by the seriousness with which Sequoia took the whole process. He found it grueling, but ultimately affirming. “There was something pretty powerful in having Sequoia come back and be like, we spoke to 20 customers. People really did love the product.” Ten days after their SF meeting, McCord had a term sheet from Sequoia in hand.

At the time of the final interview for this piece in late March 2026, the war with Iran had started just days earlier. News had just come out about the first casualties on both sides of the conflict, among them, three US soldiers. That reality was weighing on McCord for many reasons, but resonated particularly in the context of his chosen means of service and field of impact. “I’m obviously reading about those casualties and I’m thinking, could it have been prevented? Could Nominal’s technology in some way, shape, or form, have improved the hardware they were using and helped prevent their deaths? I have no idea,” says McCord.

He’s acutely aware that the state of the world has changed the way people think about hardware manufacturing, and he’s ambivalent about what it took for that shift to occur. “I don’t like that there’s a global land conflict in Ukraine and a war in Iran, but the reality is that it’s happening,” says McCord. “And I think it is pushing everyone to rethink and say, ‘Hey, building physical things is critical.”

McCord sees a silver lining to this macro shift in attention to hardware development. He’s hopeful it will enable innovations outside the realm of defense and war, and is actively expanding Nominal’s capabilities for teams building rockets and medical devices, tools for water desalination and electric vehicles. The shift is also apparent in Nominal’s rapid growth: as of May 2026, Nominal has achieved unicorn status, with 75 global customers across aerospace, defense, energy, and transportation, a rapidly growing team, and a constantly expanding product surface area with an eye towards enabling anyone to build hardware efficiently and intelligently. His team is integrating AI to further speed up data collection and analysis, and enable edge computing for systems operating beyond connectivity range—particularly essential in aeronautics and astronautics. McCord understands that service is an ongoing act, and the urgency his parents instilled in him to do something has only grown more acute with time. With each innovation, McCord returns to a mantra, one influenced by his upbringing surrounded by people inspiring him to be of service. “I call it the grandpa test. I basically ask myself all the time, ‘how will I feel when I’m old and sitting in my chair and the grandkids are around?’” says McCord. “I think I will look back very fondly if Nominal played a part in moving the physical ambitions of humanity forward. We talk about flying cars, but yes, there’s also defense, and advanced energy, and compute to power this next generation of AI. There’s advanced transportation, mobility, and water purification. There’s so much we want to do.”

AI Ascent 2026

On April 20, we hosted our fourth annual AI Ascent in San Francisco, bringing together more than 150 leading founders and researchers in AI, including Demis Hassabis, Andrej Karpathy, Greg Brockman, Boris Cherny of Anthropic, Dmitri Dolgov of Waymo, Jim Fan of Nvidia, and many more.

Sequoia partner Pat Grady opened the day with a frame for the moment: AI is a revolution in computation. Not faster horses, but cars. And the cars have arrived. His advice for founders building on top of the labs: get MAD. Build moats from the customer back, design for affordance, and exploit the diffusion gap between the model capabilities and what the Fortune 500 has deployed. Sonya Huang declared 2026 the year of agents, and walked through the three ingredients (models, tools, and harnesses) that have finally come together. Konstantine Buhler argued that the cognitive revolution will follow the same arc as the Industrial Revolution—just bigger and faster—and that AI is about to do to cognitive work what the Industrial Revolution did to manual labor.

The talks ranged from the long-horizon agent revolution and the endgame for robotics, to data centers in space, the frontier of data efficiency, and the emerging science behind neural networks.

Below is a selection of videos from the event. For the full lineup, visit our YouTube playlist.

Could pixels hold the keys to training useful agents?

The race to scale language models — and the agent ecosystem around them — is white-hot. Coding agents, which reason through problems and write code to solve them, have already taken us very far.

But one ambitious young team is making a different bet: that the most promising path to general computer agents may not run through language, screenshots, and tool calls, but through scaling raw video.

Standard Intelligence’s thesis is that the best way to build a general agent is through full video pre-training on computer use, because it is the only approach that can truly scale action data. Instead of predicting text tokens, the model learns to use a computer from raw screen data, predicting the next mouse movement, click, and keystroke from the pixels in front of it. 

It is the Tesla FSD approach applied to knowledge work on computer screens.

That makes the bet both deeply contrarian and deeply “bitter lesson”-pilled. Rather than hand-engineering workflows or wrapping language models in increasingly elaborate harnesses, Standard Intelligence is betting on a new pre-training paradigm: feed the model the raw stream of computer use, scale it aggressively, and let the generality emerge from the data.

“We’re not video people”

Video is unwieldy. It is computationally expensive, economically expensive, and technically unforgiving. Prior attempts to scale video toward AGI have often died on the vine.

The Standard Intelligence team is emphatically “not video people.” They did not arrive with a decade of inherited assumptions about how to work with video as a medium. Instead, they have had to reason through each challenge from first principles, and have met those challenges with unusual optimism, creativity, and scrappiness.

The results are striking. An 11-million-hour computer action dataset — the largest in the industry. A video encoder that is roughly 50× more token-efficient than competing approaches, enabling nearly two hours of 30 FPS video to fit inside a 1-million-token context window. A 30-petabyte storage cluster racked in San Francisco for under $500K, roughly 20× cheaper than hyperscaler alternatives.

FDM-1, their first foundation model trained directly on computer-use video at scale, offers an early glimpse of what this paradigm could become. It is a general model that can extrude a CAD gear in Blender, drive a car around a San Francisco block after an hour of fine-tuning, and find bugs in software by exploring its state space the way a curious human might.

Conscientious young founders

Founders Galen Mead and Devansh Pandey met as teenagers during the Atlas Fellowship in 2022, a selective fellowship for high-school students interested in AI alignment and AGI. 

Galen and Devansh are unusually serious about reaching AGI, and unusually conscientious about doing so safely. Both founders are wise beyond their years (21 and 20 respectively), and both left their undergraduate programs out of a sense of urgency to work on this problem.

Galen and Devansh stand out for their combination of taste, scrappiness, technical courage, and ambition. It shows up in the product thinking, in the research direction, and in the FDM-1 report itself.

The full team of six is small but mighty. Neel, Yudhister, Ulisse, and Ryan are each quirky and exceptional. They have chosen to turn down the conventional path (fancy degrees and offers from big token) and pursue this courageous mission together. 

A new pre-training regime

Video has long been a powerful training ground for AI. DQN showed that agents could learn rich behavior directly from pixels in Atari environments. Tesla scaled video models to make self-driving cars and robots navigate the physical world.

But in the race toward general knowledge agents, video-first pre-training remains an unconventional idea.

Standard Intelligence is betting that it will not stay unconventional for long.

We are thrilled to lead Standard Intelligence’s Series A alongside Miko and Yasmin from Spark Capital.

What does it mean to build a mind that is not a copy of our own?

Today, we are excited to announce that Sequoia is partnering with David Silver and Ineffable Intelligence, a new AI research lab based in London with a singular mission: to make first contact with superintelligence.

Ineffable is building what David calls a superlearner: a system that discovers all knowledge directly through its own experience, from elementary motor skills to profound intellectual breakthroughs. No pre-training. No imitation. Just an agent learning endlessly from the consequences of its own actions in a world built to teach it.

A Reinforcement Learning-based superlearner has the potential to rediscover and then transcend the greatest inventions in human history: language, science, mathematics and technology. Imagine a machine that derives the laws of physics from first principles. That invents new branches of mathematics we never thought to ask about. That designs materials, medicines and computers we don’t yet have the vocabulary to describe. This is the prize David is reaching for.

A Different Path

The current generation of AI was built by training on the entirety of the human internet. It is an extraordinary achievement. But a system trained on human data may also have fundamental limitations.

Ineffable Intelligence is scaling reinforcement learning from a clean base: no pre-training, no human data to shortcut the system. Guided by the Era of Experience as a north star, David is proving that agents trained purely from an environment can develop non-human strategies for reasoning about problems we don’t yet know how to solve.

David led some of the defining breakthroughs in deep reinforcement learning, most famously in the devilishly difficult game of Go. Go is the ultimate machine intelligence test because it cannot be brute-forced by a computer: a combinatorially explosive O(10^170) possible legal board positions vastly exceeds the O(10^80) atoms in the observable universe. It was thought to be simply too hard for machines to solve.

At DeepMind, David drove the key breakthrough that finally solved the game of Go: self-play. Self-play drove the ~800 ELO-point leap that led to the historic AlphaGo vs. Lee Sedol showdown in March 2016. David pushed the idea further still with AlphaGo Zero: removing human pre-training entirely and learning purely through self-play increased the system’s ELO rating from ~3,700 to 5,000+. The result was a system that reached decisively superhuman performance, and with somewhat non-human mannerisms.

That’s the lineage David has spent his career building. He was the lead researcher and technical force behind the Alpha series at DeepMind, where, for a brilliant period, his approach was the dominant paradigm: AlphaGo, AlphaZero, AlphaStar, AlphaProof and more.

Even with the arrival of LLMs, David never stopped believing. He is one of the very few people on earth with the conviction, the technical depth and the team to scale reinforcement learning.

What’s Ahead

The work ahead is hard, the timeline to superintelligence is uncertain, and the bet is genuinely contrarian. That is exactly what excites us. The largest leaps in AI have always come from people willing to ignore the consensus. David has ignored more consensus, more correctly, than almost anyone in the field.

We are honored to co-lead Ineffable’s first round and to partner with David on what may be the most ambitious scientific mission of our generation.

Partnering with Auctor: AI Autopilot for Software Implementation

For every dollar spent on software, six are spent on services. Why? Because buying enterprise software is easy. Implementing it is hard, and still leans heavily on labour. Across the top 10 software ecosystems (ServiceNow, Salesforce, SAP, AWS, and others), 9 million implementation consultants represent more than $500 billion in annual labour spend, still growing 10%+ a year. It’s one of the largest markets in technology, and one of the least disrupted.

Enterprise platforms have thousands of components that change daily. A single deployment can span hundreds of requirements, dozens of stakeholders, and months of back-and-forth between what a business needs and what the system can actually do. Today, consultants hold all of this together through experience and pattern recognition. But they forget context between meetings. They miss dependencies across systems. They can’t track thousands of platform updates simultaneously. Unlike LLMs, their context windows are limited by biology. 

Auctor is the autopilot for software implementation, from the moment a customer requirement is captured all the way through to delivery. It brings together the requirements, decisions, and context that typically live across meetings, documents, and dozens of disconnected systems, and translates them directly into the outputs needed to move a project forward. What used to take teams weeks of scoping now gets done in a single sitting.

We first partnered with Auctor at the Seed, shortly after the team graduated from YC. At Sequoia, each year we select a handful of portfolio companies and take their founders to meet executives at some of the most important technology companies in Silicon Valley. For an hour, these leaders effectively join the team, helping us and the founders think through strategy, positioning, and go-to-market. In our very first meeting with the C-suite of a major enterprise platform, the executives asked for a pilot halfway through the conversation, before we’d even demoed the product. 

What gave us conviction beyond the market was the team. This is a category where speed of execution determines the winner given the first mover advantage. Will, Sky, and Matt are the hardest-charging, fastest-moving team we’ve met in this space. They shipped their platform and landed their first major enterprise contract within months of graduating YC. They’ve set up shop in New York and are pulling in talent the way only a team with this much momentum can.

We’re leading Auctor’s Series A and couldn’t be more excited to partner with Will and the team to bring software implementation into the age of AI.

From Hierarchy to Intelligence

At Sequoia, we see that speed is the best predictor of start-up success. Most companies are focused on AI as a productivity enhancer. Few are focused on the potential of AI to change how we work together. Block is showing what it looks like to fundamentally rethink organization design, ultimately harnessing AI to increase speed as a compounding competitive advantage.

Two thousand years before the first corporate org chart, the Roman Army solved a problem that every large organization still faces: how do you coordinate thousands of people across vast distances with limited communication?

Their answer was a nested hierarchy with a consistent span of control at every level. The smallest unit was the contubernium, eight soldiers who shared a tent, equipment, and a mule, led by a decanus. Ten contubernia formed a century of eighty men under a centurion. Six centuries made a cohort. Ten cohorts made a legion of roughly 5,000. At each layer, a named commander held defined authority, aggregated information from below, and relayed decisions from above. The structure (8 → 80 → 480 → 5,000) was an information routing protocol built around a simple human limitation: a leader can effectively manage somewhere between three and eight people. The Romans discovered this through centuries of warfare. Even today, the US Army’s hierarchical chain follows a similar pattern. We now call it “span of control,” and it remains the governing constraint of every large organization on earth.

The next big change came from Prussia. After Napoleon’s army destroyed the Prussian forces at the Battle of Jena in 1806, a group of reformers led by Scharnhorst and Gneisenau rebuilt the military around an uncomfortable truth: you cannot depend on individual genius at the top. You need a system. They created the General Staff, a dedicated class of trained officers whose job was not to fight but to plan operations, process information, and coordinate across units. Scharnhorst intended these staff officers to “support incompetent Generals, providing the talents that might otherwise be wanting among leaders and commanders.” This was middle management before the term existed. Professionals whose purpose was to route information, pre-compute decisions, and maintain alignment across a complex organization. The military also formalized the distinction between “line” and “staff” functions. Line advances the core mission. Staff provides specialized support. Every corporation still uses this vocabulary today.

Military hierarchy entered the business world through the American railroads in the 1840s and 1850s. The U.S. Army lent West Point-trained engineers to private railroad companies, and these officers brought military organizational thinking with them. Staff and line hierarchies, divisional structure, bureaucratic systems of reporting and control: all of it was developed in the military before the railroads adopted it. In the mid-1850s, Daniel McCallum of the New York and Erie Railroad created the world’s first organizational chart to manage a system stretching over 500 miles with thousands of workers. The informal management styles that worked for smaller railroads were failing. Train collisions were killing people. McCallum’s chart formalized the same hierarchical logic the Romans had used: layers of authority, defined reporting lines, structured information flow. It became the blueprint for the modern corporation.

Frederick Taylor (1856-1915), often called the “Father of Scientific Management,” optimized what happened within that hierarchy. Taylor broke work into specialized tasks, assigned them to trained experts, and managed through measurement rather than intuition. This produced the functional pyramid organization – a structure optimized for efficiency within the information routing system that the military had pioneered and the railroads had commercialized.

The first real stress test of functional hierarchy came during World War II. The Manhattan Project required physicists, chemists, engineers, metallurgists, and military officers to work across disciplinary boundaries toward a single objective under extreme secrecy and time pressure. Robert Oppenheimer organized Los Alamos into functional divisions but insisted on open collaboration across them, resisting the military’s instinct to compartmentalize. When the implosion problem became critical in 1944, he reorganized the lab around it, creating cross-functional teams unlike anything in corporate America at the time. It worked, but it was a wartime exception led by a singular figure. The question the postwar business world faced was whether that kind of cross-functional coordination could be made routine.

With the growth and globalization of companies after World War II, the scale limitations of functional design became acute. In 1959, McKinsey’s Gilbert Clee and Alfred di Scipio published “Creating a World Enterprise” in the Harvard Business Review, providing an intellectual framework for a matrix organization that combined functional specialties with divisional units. Under the leadership of Marvin Bower, McKinsey helped companies like Shell and GE implement these principles, balancing central standards with local agility. This became the “professional” or “modern” corporation that propelled the postwar global economy.

Over time, other frameworks emerged to address the complexity, rigidity, and bureaucracy of matrix structures. The McKinsey 7-S framework, developed in the late 1970s by Tom Peters and Robert Waterman, distinguished the “hard Ss” (Strategy, Structure, Systems) from the “soft Ss” (Shared Values, Skills, Staff, Style). The core idea was that structural elements alone were insufficient. Organizational effectiveness required alignment across cultural traits and the human factors that determine whether a strategy actually succeeds.

In more recent decades, technology companies have experimented aggressively with organization structure. Spotify popularized cross-functional squads with short sprint cycles. Zappos attempted Holacracy, eliminating management titles entirely. Valve operated with a flat structure and no formal hierarchy. Each of these experiments revealed something about the limitations of traditional hierarchy, but none solved the underlying problem. Spotify moved back toward conventional management as it scaled. Zappos saw significant attrition. Valve’s model proved difficult to scale beyond a few hundred people. As organizations grow into the thousands, they revert to hierarchical coordination because no alternative information routing mechanism has been powerful enough to replace it.

The constraint is the same one the Romans faced and the Marine Corps rediscovered in World War II: narrowing span of control means adding layers of command, but more layers mean slower information flow. Two thousand years of organizational innovation has been an attempt to work around this tradeoff without breaking it.

So what’s different now?

At Block, we’re questioning the underlying assumption: that organizations have to be hierarchically organized with humans as the coordination mechanism. Instead, we intend to replace what the hierarchy does. Most companies using AI today are giving everyone a copilot, which makes the existing structure work slightly better without changing it. We’re after something different: a company built as an intelligence (or mini-AGI).

We are not the first to try to move beyond traditional hierarchy. Haier’s rendanheyi model, platform organizations, “data-driven” management: these are real attempts at the same problem. What they lacked was a technology capable of actually performing the coordination functions that hierarchy exists to provide. AI is that technology. For the first time, a system can maintain a continuously updated model of an entire business and use it to coordinate work in ways that previously required humans relaying information through layers of management.

For this to work, a company needs two things: a kind of “world model” of its own operations, and a customer signal rich enough to make that model useful.

Block is remote-first. Everything we do creates artifacts. Decisions, discussions, code, designs, plans, problems, and progress all exist as recorded actions. It’s the raw material for a company world model. In a traditional company, a manager’s job is to know what’s happening across their team and relay that context up and down the chain. In a remote-first company where work is already machine-readable, AI can build and maintain that picture continuously. What’s being built, what’s blocked, where resources are allocated, what’s working and what isn’t. That’s the information the hierarchy used to carry. The company world model carries it instead.

But the capability of the system is only as good as the quality of the customer signal feeding it. And money is the most honest signal in the world.

People lie on surveys. They ignore ads. They abandon carts. But when they spend, save, send, borrow, or repay, that’s the truth. Every transaction is a fact about someone’s life. Block sees both sides of millions of these transactions every day, the buyer through Cash App and the seller through Square, plus the operational data from running the merchant’s business. That gives the customer world model something rare: a per-customer, per-merchant understanding of financial reality built from honest signal that compounds. The richer the signal, the better the model. The better the model, the more transactions. The more transactions, the richer the signal.

Together, the company world model and the customer world model form the foundation for a different kind of company. Instead of product teams building predetermined roadmaps, you build four things.

First, capabilities. The atomic financial primitives: payments, lending, card issuance, banking, buy-now-pay-later, payroll, and so on. These are not products. They are building blocks that are hard to acquire and maintain (some have network effects and regulatory permission). They have no UIs of their own. They have reliability, compliance, and performance targets.

Second, a world model. This has two sides. The company world model is how the company understands itself and its own operations, performance, and priorities, replacing the information that used to flow through layers of management. The customer world model is the per-customer, per-merchant, per-market representation built from proprietary transaction data. It starts with raw transaction data today and evolves toward full causal and predictive models over time.

Third, an intelligence layer. This is what composes capabilities into solutions for specific customers at specific moments and delivers them proactively. A restaurant’s cash flow is tightening ahead of a seasonal dip the model has seen before. The intelligence layer composes a short-term loan from the lending capability, adjusts the repayment schedule using the payments capability, and surfaces it to the merchant before they even think to look for financing. A Cash App user’s spending pattern shifts in a way the model associates with a move to a new city. The intelligence layer composes a new direct deposit setup, a Cash App Card with boosted categories for their new neighborhood, and a savings goal calibrated to their updated income. No product manager decided to build either solution. The capabilities existed. The intelligence layer recognized the moment and composed them.

Fourth, interfaces (hardware and software). Square, Cash App, Afterpay, TIDAL, bitkey, proto. These are delivery surfaces through which the intelligence layer delivers composed solutions. They are important, but they are not where the value is created. The value is in the model and the intelligence.

When the intelligence layer tries to compose a solution and can’t because the capability doesn’t exist, that failure signal is the future roadmap. The traditional roadmap, where product managers hypothesize about what to build next, is any company’s ultimate limiting factor. In this model, customer reality generates the backlog directly.

If this is what the company builds, then the question becomes: what do the people do?

The org structure follows from this, and it inverts the traditional picture. In a conventional company, the intelligence is spread throughout the people and the hierarchy routes it. In this model, the intelligence lives in the system. The people are on the edge. The edge is where the action is.

The edge is where the intelligence makes contact with reality. People reach into places the model can’t go yet. They sense things the model can’t perceive: intuition, opinionated direction, cultural context, trust dynamics, the feeling in a room. They make the calls the model shouldn’t make on its own, especially ethical decisions, novel situations, and high-stakes moments where the cost of being wrong is existential. A world model that can’t touch the world is just a database. But the edge doesn’t need layers of management to coordinate it. The world model gives every person at the edge the context they need to act without waiting for information to travel up and down a chain of command.

In practice, this means we normalize down to three roles.

Individual contributors (ICs) who build and operate capabilities, the model, the intelligence layer, and the interfaces. They are deep specialists and experts in a specific layer of the system. The world model provides the context that a manager used to provide, so ICs can make decisions about their layer without waiting to be told what to do.

Directly Responsible Individuals (DRI) who own specific cross-cutting problems or opportunities and customer outcomes. A DRI might own the problem of merchant churn in a specific segment for 90 days, with full authority to pull resources from the world model team, the lending capability team, and the interface team as needed. DRIs may persist on certain problems or move elsewhere to solve new ones.

Player-coaches who combine building with developing people. They replace the traditional manager whose primary job was information routing. A player-coach still writes code or builds models or designs interfaces. They also invest in the growth of the people around them. They don’t spend their days in status meetings, alignment sessions, and priority negotiations. The world model handles alignment. The DRI structure handles strategy and priority. The player-coach handles craft and people.

There is no need for a permanent middle management layer. Everything else the old hierarchy did, the system coordinates, and everyone is empowered, with a role that’s much closer to the work and the customer.

Block is in the early stages of this transition. It will be a difficult one, and parts of it will likely break before they work. We’re writing about it now because we believe every company will eventually need to confront the same question we did: what does your company understand that is genuinely hard to understand, and is that understanding getting deeper every day?

If the answer is nothing, AI is just a cost optimization story. You cut headcount, improve margins for a few quarters, and eventually get absorbed by something smarter. If the answer is deep, AI doesn’t augment your company. It reveals what your company actually is.

Block’s answer is the economic graph: millions of merchants and consumers, both sides of every transaction, financial behavior observed in real time. That understanding compounds every second the system operates. We believe the pattern behind this, a company organized as an intelligence rather than a hierarchy, is significant enough that it will reshape how companies of all kinds operate over the coming years. Block is far enough along to show the idea is more than theory (though, we welcome debate and feedback to pressure test and improve our ideas).

Companies move fast or slow based on information flow. Hierarchy and middle management impede information flow. For two thousand years, from the Roman contubernium to today’s global enterprises, we have had no real alternative. Eight soldiers sharing a tent needed a decanus. Eighty men needed a centurion. Five thousand needed a legate. The question was never whether you needed layers. The question was whether humans were the only option for what those layers do. They aren’t anymore. Block is building what comes next.

Partnering with Edra: Context for Agents at Scale

Every company runs differently. Two businesses in the same industry will have their own escalation paths, and workarounds—their own tribal knowledge accumulated over years and stored, if anywhere, in the institutional memory of people who won’t always be there. When you drop a general-purpose AI into that environment, it starts from zero. The work of getting it up to speed (the forward-deployed engineers, manual documentation, consultants) is slow, expensive and has to be redone every time a process changes. Most companies are living this problem right now.

Eugen Alpeza spent seven years at Palantir, where he was instrumental in building the company’s U.S. commercial go-to-market motion, including starting Palantir’s work with AT&T, one of its largest and most complex deployments. In 2023, he took on the launch of Palantir’s AI Platform under CEO Alex Karp. Together with Yannis Karamanlakis, they created the Forward Deployed AI Engineer role at Palantir—designed to bridge AI research with real world production deployments. Yannis became the first Forward Deployed AI Engineer at the company, leading a team focused on taking LLMs from demos into production at scale. Yannis had already led a major pure AI commercial project, a recruiting search engine that increased placement rates for a staffing firm by 129%. The two left Palantir as close friends and co-founders. They had known each other for 13 years, since university, and had long planned to start a company together.

What Edra has built is elegant in its logic. Instead of asking humans to document processes, Edra analyzes the data a company already generates. Through support tickets, emails, logs, chat histories, it creates a living knowledge base that reflects how the business actually runs, not just how it was supposed to run on paper. As people use it, the system learns and improves on its own. And unlike black-box fine-tuning approaches, it is transparent and editable—you can see exactly what Edra has learned and why. From there, agent automation is straightforward. 

The early results are real. The first successful use cases are around automating IT service management and customer technical support, where the data is rich and the pain is acute. The early customers love it and are expanding aggressively. 

As always, our investments are all about people. When I first met Eugen and Yannis, what struck me was not only what they had built, but how they work together. Eugen is one of the most commercially gifted people I have met—someone who earns the trust of skeptical buyers and makes them believe. Yannis is technically exceptional, the kind of partner who makes the hardest things feel solid. Their dynamic as a founding duo is a genuine superpower.

We are thrilled to partner with Eugen, Yannis and the entire Edra team. 

Partnering with Scanner: Every Log Tells a Story—If You Can Find It Fast Enough

A while back, I was deep in research on the next generation of security infrastructure, talking to CISOs and security engineers at some of the most technically sophisticated companies in Silicon Valley. I asked them all the same question I’d asked a decade earlier when I worked in enterprise software: What’s your biggest headache? The consistency of their answers surprised me. “We drown in logs we can’t afford to keep,” as one security leader put it, “and go blind on the logs we can’t afford to search.”

Enterprise security today is a story of impossible choices. The tools that teams rely on generate enormous amounts of log data—every API call, every login event, every network connection. To investigate cyber threats, they need all of it, often going back a year or more. But storing everything in a SIEM like Splunk is prohibitively expensive; costs could easily consume 15% of a CISO’s entire budget. Instead, companies make a compromise: they keep only the most recent 10 to 30 days of logs in their SIEM and park the rest in Amazon S3, where storage is cheap, but the data is effectively frozen. When a breach, a compliance audit, or a forensic investigation happens, security teams discover too late that the evidence they need is out of reach, opaque and unsearchable. 

I first heard about Scanner from a member of the security team at Temporal, one of our portfolio companies, who called it, “crazy fast.” I looked into it, and reached out to Cliff Crosland right away.

What Cliff and his co-founder Steven Wu have built is elegant in its insight. They asked: what would a log search engine look like if you designed it from scratch for object storage? The answer was a purpose-built inverted index that maps field values directly to file regions in S3. Rather than combing through billions of rows, Scanner narrows each query to only the relevant slices of data. A petabyte of logs becomes interactive. Queries that took hours now run in seconds. And a streaming detection engine runs hundreds of detection rules continuously across tens of terabytes a day, without re-scanning the world for each one.

Cliff and Steven are exactly the kind of founders we look for. Both Stanford CS alums, they were engineering leads together at Accompany (acquired by Cisco), where they built core data infrastructure under demanding, production-scale conditions. They have an obsession with performance that borders on the philosophical; they don’t tolerate systems that feel slow. And they have the expertise to build something better.

What’s most striking about Scanner isn’t the technology—though that is genuinely impressive. It’s the customers. The companies using Scanner today read like a who’s who of the cloud native world: Notion, Ramp, Benchling, Confluent, Lemonade, BeyondTrust. And they’re not just using it—they love it. Benchling replaced another product after a forced tenfold price increase, and their head of security engineering called it one of the best technical decisions their team had made. Ramp started with security logs and then expanded to application logs, reducing their SIEM bill in the process. Notion’s detection and response team built an internal AI agent that autonomously runs security investigations using Scanner. 

That last example signals what’s to come. We are entering a new era of security operations, where AI agents will do much of the investigative work that today consumes hours of human time. But agents need to rapidly iterate, ask questions and follow threads; queries can’t take minutes, much less hours. Scanner’s speed is enabling these agentic security workflows across a wide range of companies: within weeks of their MCP release, nearly a third of Scanner’s customers were already using it in production, and agents now account for 80% of queries on the platform. That is not a prototype or a promising beta. That is the future arriving ahead of schedule.

Sequoia is proud to lead Scanner’s Series A, and we’re thrilled to partner with Cliff, Steven and their team as they work to transform a market overdue for reinvention. Scanner is winning hearts and minds among the most technically forward organizations today, and together, they will define the next decade of security infrastructure.