Sequoia AI Ascent - AI might be the biggest opportunity in humanity’s history
Sequoia’s AI Playbook: The Agent Economy, Application Layer, and How to Win in 2025
Hey everyone, my name is Guillermo Flor and I’m an entrepreneur and Venture Capitalist.
Last week Sequoia hosted AI Ascent, a private event that brought together some of the sharpest minds in artificial intelligence to answer the biggest questions in the field today.
I just knew I had to study in depth and share it with you guys. Subscribe if you found it valuable and share with other founders.
We are living through the single greatest entrepreneurial opportunity of our lifetime—and maybe in human history.
At AI Ascent, Sequoia brought together some of the sharpest minds in AI to answer the questions that matter most:
What’s real? What’s hype? Where’s the value going? And—most importantly—how do we build the future?
Here’s the full breakdown of AI Ascent:
Why AI is starting 10x bigger than the cloud boom
How distribution has changed—and why AI spread faster than anything before
Where the real value will be captured (hint: it’s not the models)
How to actually win: the 95% that’s timeless and the 5% that’s AI-specific
Sequoia’s checklist for evaluating AI companies: revenue, margins, moats
The Vacuum Effect: Why You Must Move Fast
The Agent Opportunity in 2025 🔥
8. What does the future look like? Welcome to the Agent Economy
1. Why AI is starting 10x bigger than the cloud boom
Sequoia opened with a simple framework: What is AI? Why does it matter? Why now? And what should we do about it?
The key point: this shift is bigger than the cloud transition.
Back then, software was a $400B market. AI is going after software and services at the same time—both huge markets. And the best companies are moving fast: from building tools, to co-pilots, to full systems that can handle work end-to-end.
2. How distribution has changed—and why AI spread faster than anything before
To go viral, a product needs:
Awareness
Desire
Accessibility
When Salesforce started, it had to fight just to get people’s attention.
Back then, only 200 million people were online.
Now it’s 5.6 billion. Everyone’s connected. Platforms like Reddit, X, and TikTok make distribution instant.
When ChatGPT launched, the world was already listening. The rails were in place. That’s why it exploded. And this isn’t unique to OpenAI—this is the new normal.
If your product’s good, it can go global overnight.
3. Where the real value will be captured (hint: it’s not the models)
Yes, the models are powerful. But the value? It’s still in the apps.
Sequoia believes most of the value will sit at the application layer—not in the infrastructure, but in the software people actually use. That’s how it played out with cloud and mobile, and the same pattern is taking shape in AI.
But it’s competitive. Foundation models aren’t staying in their lane—they’re moving into application territory too. That means startups need to build from the customer back. Focus on a specific user, in a specific market, with a real workflow.
Go deep, not wide. That’s where startups can still win.
These logos represent the companies that got to 1b in revenues: most of them are at the application level 👇
4. How to Build a Real AI Company
How do we play to win: 95% of building company in ai is jsut building a company.
The other 5% is the ai: this is what you need to take an idea and turn it into a product
The other 5% is AI specific: Dog Leoni's merchandising cicle 👇
1. Your customer doesn't know how to use AI: you can have an opinion (Vision)
2. You build an end to end solution to their problem (Product)
3. You can build data flyhweels with the usage data of your own product (Engineering)
4. You can be of the industry of the industry (Marketing)
5. You can speak their language (Sales)
6. You can have great customer support (Support)
5. Sequoia’s checklist for evaluating AI companies: revenue, margins, moats
Real Revenue vs. Vibe Revenue
Don’t get high on vibes. Inspect retention, adoption, and behavior change.
Gross Margins
Your margins may suck today. That’s okay—token costs are dropping fast.
Are you moving from tool → outcome → strategic partner? That’s where the money is.
Data Flywheels
Ask yourself: what business metric does my data flywheel improve?
If you don’t know, it’s probably not real.
6. The Vacuum Effect: Why You Must Move Fast
There’s a huge gap in the market. Everyone knows AI is coming, but most industries aren’t ready for it. That creates space—and whoever gets there first will win.
Ignore the noise. Macro conditions won’t stop this. AI is happening either way. If you don’t move now, someone else will. It’s that simple.
7. The Agent Opportunity in 2025 🔥
Last year, AI apps had weak engagement. This year, usage looks more like social media. ChatGPT is closing in on Reddit levels of daily use. AI is no longer a novelty—it’s becoming part of daily life.
And this goes beyond fun tools or cool demos. Startups are using AI to write better ads. Teachers are using it to create lesson plans. Doctors are using it to check diagnoses. This stuff is working—and it’s only just getting started.
8. What does the future look like? Welcome to the Agent Economy
We’re moving fast from individual AI tools to full networks of agents. These agents don’t just respond to commands—they collaborate, reason, and complete tasks together. That’s what Sequoia calls agent swarms.
The next step is the agent economy: a world where agents don’t just talk to each other—they trade value, track trust, and take action. They act like participants in a real market.
This doesn’t replace humans. It gives us more leverage. A solo founder can do the work of a 20-person team. A lean company can move like a giant. That’s the kind of shift we’re walking into.
The 3 hardest technical problems left to solve (identity, communication, security)
There are still some big technical hurdles before this future is fully live:
Agents need memory—they have to remember you, and themselves.
They need communication protocols—the equivalent of TCP/IP for AI.
And we need new forms of security and trust, because you won’t meet your agent face-to-face.
Solving these problems will unlock a whole new layer of capability.
The Stochastic Mindset
This shift also requires a new mindset. Sequoia called it the stochastic mindset—a way of thinking that accepts that not everything will be predictable or repeatable. That’s the nature of AI. You’re managing systems that think, not scripts that execute.
You’ll also need a management mindset. You’re not just writing code—you’re managing agents. Giving feedback. Debugging workflows. Making decisions about how they should act and what they should do.
And more than anything, you’ll need to be comfortable with high leverage and high uncertainty. You can get more done than ever before—but you won’t always know exactly how it’s happening. That’s the tradeoff.
🔍 About This Article
This piece is part of The AI Opportunity, a newsletter read by founders, VCs, and AI operators who want signal, not noise. In this edition, we break down Sequoia Capital’s AI Ascent 2025—a private event where top Sequoia partners revealed how they see the future of AI:
Where the agent economy is heading
What’s hype vs real in the AI startup landscape
How to build enduring value at the application layer
What metrics matter when building AI products
And why this moment is the biggest tech opportunity since the internet
These FAQs are designed to help you—and the search engines—quickly understand the key ideas in the article and why they matter.
❓FAQs — Sequoia’s AI Strategy, The Agent Economy, and Building in the Age of AI
What is Sequoia’s AI Ascent?
AI Ascent is a private event hosted by Sequoia Capital where partners like Pat Grady, Sonia Verma, and Constantine Valhouli share their latest thinking on the AI market. The 2025 edition covered foundational shifts in the AI stack, the rise of the agent economy, and how Sequoia evaluates AI startups.
What is the agent economy in AI?
The agent economy is the emerging system where autonomous AI agents collaborate, transact, and solve complex problems across industries. These agents operate as economic participants—doing work, communicating with other agents, and creating value with minimal human input. Sequoia sees this as the next frontier in AI.
Where does Sequoia believe the value will accrue in AI?
Sequoia believes most of the value will be captured at the application layer—not in the foundation models themselves. This means vertical tools, industry-specific copilots, and software that solves real problems will outperform general-purpose infrastructure in terms of revenue and defensibility.
How should startups think about building in AI today?
Sequoia urges founders to go vertical and customer-back. Rather than competing with foundation models, build AI-native applications for specific workflows. Look for places where AI moves from “tool” to “co-pilot” to “autopilot”—and where you can collect proprietary data to improve the product over time.
What is a data flywheel in AI, and why is it important?
A data flywheel happens when your AI product generates usage data that improves the model and experience for future users. This compounds over time, creating a defensible moat. Sequoia emphasizes that a flywheel only matters if it moves a real business metric like retention, revenue, or performance.
What are Sequoia’s criteria for investing in AI startups?
Sequoia looks at the same fundamentals as always—team, market, product—but with AI-specific filters:
Is the revenue real, or just “vibe revenue”?
Are margins trending in the right direction as token costs drop?
Does the product generate a data flywheel that compounds over time?
They also look for trust-building with users, which is critical in early-stage AI adoption.
What is the biggest opportunity in AI right now?
According to Sequoia, the biggest near-term opportunity lies in vertical AI applications—especially those that integrate agents into real workflows. Startups solving domain-specific problems with high frequency, high stakes, and messy data will have the best shot at building lasting value.
How do AI agents differ from traditional automation?
AI agents go beyond automation by reasoning, adapting, and working with other agents. Instead of hard-coded instructions, agents can interpret intent, make decisions, and even negotiate with other systems. This creates more flexible, scalable workflows—especially in industries like security, DevOps, healthcare, and finance.
What makes The AI Opportunity newsletter different?
The AI Opportunity focuses on what actually matters in AI—founder strategy, GTM playbooks, emerging products, and how VCs like Sequoia, a16z, and Index are thinking. It’s written by someone inside the ecosystem, not a trend-chaser.
Cheers,
Guillermo
A very interesting article. Thank you very much. Summed up the current trend of AI in a beautiful and forward looking way.
Interesting read. Thank you!