👋 Howdy to the 4,003 new legends who joined since our last edition! You are now part of a 446,555-strong tribe outperforming the competition together.

LATEST POSTS 📚

If you’re new, not yet a subscriber, or just plain missed it, here are some of our recent editions.

🎳 Brand Expansion = Loading. We've reached a point to take a breath and think of what should come next.
🧬 Enhancement, AI Doctors & The Future Of Health. An interview with Max Marchione, Founder at Superpower.
🔥 How To (Attempt To) Send 1M Outbound Emails A Day. Our wild plan to send world record levels of outbound.

PARTNERS 💫

Framer helps teams design, build, and launch their marketing sites lightning fast. With the ability to publish hundreds of CMS pages in a single click, operate at a global scale with seamless localization, and even host unified content across multiple domains, teams have never been able to ship faster.

Trusted by companies like Miro, Bilt, and Perplexity.

For Small Business Month, get $1,500+ savings to back your next business move. Fuel your business on America's best network and score $100 back per voice line on T-Mobile for Business when you add 15+ lines SuperMobile plan.*

Get the best for what's ahead with T-Mobile for Business' most secure, high-performance communication solution—with no annual contracts. Switch now and supercharge your business with SuperMobile.

^{*With 10 monthly bill credits. 15+ lines of eligible service ($45+/mo. w/ AutoPay; plus taxes & fees) required.}

_{Interested in sponsoring these emails? See our partnership options here.}

HOUSEKEEPING 📨

If you haven’t noticed, there are some pretty interesting plays on the agent harness today. They are very interesting to me because being trapped in one ecosystem (Anthropic, OpenAI, others) while other models accelerate around you is stressful. The team at Hyperagent have built a platform where you use whichever model is best for the job. Claude, ChatPT, Gemini.

They also gave me $1,000 free credits for my community to try it out. Hyperagent is a new startup inside of Airtable, the $11B behemoth. So quality is all over the tool. Anyway, let’s dive into how they built it and the strategy around it.

INTERVIEW 🎙️

Andrew Busse, VP of AI Operations at Airtable

Andrew Busse is the VP of AI Operations at Airtable, where he leads operations and go-to-market for Hyperagent, the company's standalone platform for building and deploying autonomous AI agents. A mechanical engineering and business management graduate of MIT, Andrew traded equities at Goldman Sachs and founded a consumer supplements brand, Revivo Energy Chews, before joining Airtable in 2018 and moving through growth, special projects, and chief of staff roles. He is also a Scout at Andreessen Horowitz. Airtable, founded in 2012 and valued at $11.7 billion, is used by more than 500,000 organizations. Hyperagent launched publicly in February 2026 with a mission to help companies go agent-first; recently, they gave away $10M to founders who are building agentic businesses. They call it the Founding 500.

The bet underneath Hyperagent is a familiar one for Airtable. A decade ago, the company took building software and made it something anyone could do. Andrew thinks the same opportunity exists with AI agents, a proven technology still locked behind technical setup and security risk, waiting for someone to make it usable by an entire team rather than a handful of power users. His argument is that the model itself is becoming a commodity, and the durable product is the layer around it—the skills, the memory, and the verification loop that improves with every run. The clearest proof so far might be their recent launch: the Founding 500 campaign that introduced Hyperagent, from founder and influencer outreach to the landing page to lead qualification, was run almost entirely by Hyperagent's own agents.

Why does Airtable need an agentic product right now?

The form factor behind Hyperagent is the autonomous agent. Give a model a real computer, real tools, and let it do the work. That got validated over the last year by OpenClaw and others. It clearly works and resonates in the market, but it's still reserved for power users. You have to be pretty technical, run it on your own machine, and accept the security risk that comes with it. The opportunity is taking that proven form factor and making it something anyone can actually run a business on. That's our bet.

Why build Hyperagent adjacent to the core product?

In terms of how we operate, it has to move at frontier speed. The market is barely formed and moving far faster than traditional software, so it needs to be unconstrained. It doesn't fit within the existing paradigm. From a development standpoint, we needed to be in an unconstrained world, but conceptually it's also a bit different. Airtable is just one of many enterprise products that Hyperagent can connect to in order to get work done. Similar to being model-agnostic, Hyperagent integrates with many enterprise tools, which means we're not going to be overly narrow in our focus on Airtable-specific use cases, though we are seeing many early users using Airtable as the data layer for their agents.

On brand, people have very limited shelf space for what they associate with a brand. The moment Airtable becomes three or four different things, it becomes nothing. Having a clear sandbox to play in and a clear definition of what this product is (so there's name recall and a strong first impression) really matters when you have a whole new form factor, as opposed to just a feature within an existing product.

What was the founding insight?

In a nutshell, AI intelligence is the prerequisite, but not the product. The real product is becoming the harness around the agent. Once models got good enough, especially around the end of last year with Claude Opus 4.6, something shifted. The bottleneck moved from just having a smart model to what you actually apply that model to.

Concretely, that's three things. Skills, where the agent learns how you actually get work done. Memories and retrieval, so it retains your context across sessions. And evals, meaning verification, where every run gets scored, and the whole thing gets better with every single run. The harness around models that are always growing and improving is the durable part. That was the founding inspiration.

How important is model agnosticism to what you're building?

Similar to how Airtable took a model-agnostic approach for its own AI features, what we see is that users have very different applications for their jobs to be done, and real preferences around models based on performance versus cost. There's an ability to truly tailor workflows and agents to specific models.

What's most exciting about AI agents for early-stage builders?

The scaling IC Founder is taking on a new life in this world. What Hyperagent really allows you to do is create an augmented fleet of agents across a lot of these workflows, such as early outreach, understanding funding rounds, competitive intel, and aggregating feedback from early customers. All those nice-to-haves that founders typically let fall through the cracks when they're focused on product development.

But the more important shift is moving from thinking of AI agents as a time-saver for all the stuff you don't want to do, to actually injecting them directly into the core of your business. That could be actual product development, integrating with your repo, creating a dynamic ship log, taking support feedback, and directly informing what you're building on your roadmap. The people having the most impact from our early customers are those taking end-to-end workflows and reimagining each step with an AI agent. Maybe you take in new clients in a totally agentic way: something new comes in, it gets assessed and qualified, a contract write-up gets generated, and ships back to the client, and the agent is actively waiting for the email reply. That's their entire business supply chain, built agentically.

How do you picture the org chart in an agentic world?

One way I like to think about it is that those closest to the problem are probably best equipped to solve it. In the old world, you'd have to file an IT ticket or go find software that might solve whatever specific thing you were experiencing. We're trying to reduce the barrier to entry. Anyone with the will, the skill, and the understanding of the business processes should be able to build, develop, and use agents.

But, as with Airtable, what we end up seeing is power concentrated among the technically capable individuals who build it into a proper org structure. A lot of what we're investing in is things like a command center, where a single person—say, your ops lead—thinks holistically about skills across the company and whether the data infrastructure is actually connected in the right way for agents to use. So yes, there will always be a low floor for people to interact with agents intuitively. But the actual configuration is probably more concentrated in these power users who lead the charge on what agents mean in a business setting.

 Who's the dream user of Hyperagent?

There are kind of two groups at this stage. The first is the wave of founders thinking in an agent-first way; a new breed asking how they can scale themselves, how to build the first one-person billion-dollar company. All that hype that exists right now. They're making a lot of sacrifices with the tools they're currently using. With OpenClaw, it's essentially YOLO mode across all your software. The question is how to do this reliably and scale, especially as the team grows. That's very much the founder looking to outmaneuver competitors. A lot of people we're seeing are, like, local tax attorneys in the San Antonio region, where, in their market, going agent-first means completely outperforming their competition because it scales them by orders of magnitude beyond traditional businesses.

The second group is existing operators at mid-market to slightly larger companies going through that same transformation, but from a very different starting point. They're coming at it from an enterprise readiness angle. That's a really interesting piece for us, especially as we build out more team functionality and governance. The founders are leading the way, and that's the core ICP, but the existing organizations looking to find leverage and efficiency are also a key part of where we're headed.

What was the biggest trade-off when shipping version 1?

Being feature maximalist. We intentionally started with a very broad set of capabilities out of the box. Memories, evals, multiple invocation types. It's powerful, but it's a lot of product surface area with implications for both UX, where there isn't always a clear golden path, and execution focus for the team.

The tension is ‘if it's too narrow, it's just a toy. If it's too broad, it's overwhelming.’ Where we landed is giving power users escape hatches. The simple paths stay simple, but the depth and control are there when you want them. That's the low-floor, high-ceiling bet. The hard part was holding both of those things true at the same time.

What does your day-to-day look like?

I've had every ambiguous title that's ever existed: growth, special projects, chief of staff, AI ops. You have to get comfortable with ambiguity. My focus now is on this new product area. Being the operational glue, so clarity on what we're building, why, and by whom, plus all the systems that enable that. But it's also owning the go-to-market side.

The team is small but mighty. There's a heavy focus on agentically leveraged developers. On the go-to-market side, it's about being as scrappy as possible and rethinking go-to-market the way a startup would. Our asymmetric advantage is Airtable's balance sheet, but a lot of what I'm managing is figuring out which growth bets to make and how to bring this vision to life with its own brand identity. A lot of my week is either building and perfecting agents, working on go-to-market and the campaigns we're driving, or closing the loop with the development teams: ‘What are we hearing from the market, are we still focused on the right big rocks, and where does that take the vision?’

Where does the Hyperagent team sit in Airtable's org chart?

On the go-to-market side, I'm running daily standups, working with the marketing team, managing ops. On the dev side, Howie is driving the roadmap and making sure we're executing against the right vision. Howie is one of the best examples of the IC CEO in practice, pushing more code and burning more tokens than probably anyone else on the team. It's a very invested-in bet. The failure mode in a big company is treating it like a pet project, having it buried in a team within a team within a team. Fin would have died if that's how Intercom had handled it. It has to be elevated in a way that gets the energy and attention required to get it over the zero-to-one hump.

As a team, we try to dedicate as much time as possible to avoid context switching; what I'd call a strategy tax. In my role, keeping the operating rhythm going, forming a clear vision, getting the right people on the right things; it's a full-time job focused on Hyperagent.

What does the near-term roadmap look like?

If I think of the pyramid of product needs, the foundation is the agent harness, the skills, memories, and memory retrieval. Then you get into the more differentiated features on top. One big near-term focus is team features. Today you can build an agent, deploy it to Slack, and invoke it, but we want to get to a world where the product truly meets the demands of teams. Collaborating on agents together, which comes with permissions, cost, and billing implications. The core near-term goal is making Hyperagent the place teams go to deploy agents safely, reliably, and performantly, with full governance and visibility into their organizations. Beyond that, more model agnosticism, getting more models into the platform so we can meet users where they are.

Why isn't collaboration a built-in feature of the existing model providers?

It comes with a lot of considerations you have to be really thoughtful about. The way most agents work today is by authenticating into systems on someone's behalf. So there are major permissions questions. If I connect to my Databricks instance, or I'm on the HR team and connect to Workday, all of a sudden my agent has access to compensation data that someone else working on that agent shouldn't have.

When the scope is agent-to-user, permissioning and governance are more manageable. When you get into multiplayer, there are harder questions. If an agent is authenticating into someone else's account, can that person change the system prompt? Can they deploy it somewhere else?

What Airtable has done well historically, and why we have confidence in the people building this product, is taking these complex software decisions around permission architecture and team sharing and turning them into simple, tasteful UX. You could probably hack that together right now with an OpenClaw instance, but it'll be ugly and have holes. Building something a team can actually adopt and understand. That's where the magic is, and that's what we're building toward.

💡Reminder: Hyperagent is giving away $1,000 in credits to the first 1,000 OpenSource CEO subscribers. Claim yours now.

What's your favorite agentic workflow day-to-day?

I have two at the moment, and I'll admit the first is a simpler one. The workflow that has saved me the most time is our data analyst agent. I used to spend a lot of time writing SQL. I know it sounds like something you could just use a chatbot for, but the way we've structured this (with skills, a repo of cascading skill retrieval, and specific definitions for things like how we calculate MRR), we've built a really robust data analyst that produces better analysis than I ever did manually. We had our board meeting today, and I was able to get full analyst-level questions answered on the fly in a way that a model alone couldn't do. The skills, the memories, the ability to share it with the data team. That's a real unlock.

How do you get the best out of yourself, personally and professionally?

On the work side, being maniacally focused on the highest-leverage thing at any given point, and treating delegation as a feature, not a failure mode. That second part was the harder thing to learn. My instinct, like probably a lot of founders, is to just take on everything myself. Getting the best out of me is really about fighting that instinct and applying myself like a laser rather than a floodlight.

On the personal side, reset moments. Staying active, going to the gym, hiking, getting off the screen. I spend so much of my life staring at a screen. Cycling is the big one. Last year I was aiming to ride 100 miles every single week, which is hard to pull off with a new kid. We just went to Yosemite this past weekend, and coming back from that is exactly where I find I can really reset and clear the noise.

Extra reading

• Building Athyna Intelligence - January, 2026

• AGI is here. Now harness it. - February, 2026

• The World’s Hottest Industry: Post-Training - March, 2026

And that’s it! You can follow Andrew on LinkedIn or check out Hyperagent on their website to keep up with what they’re building!

BRAIN FOOD 🧠 

TWEETS OF THE WEEK 🐣 

TOOLS WE RECOMMEND 🛠️

Every week, we highlight tools we like and those we actually use inside our business and give them an honest review. Today, we are highlighting Lightfield*—an AI-native CRM that assembles itself from your email, calendar, and meetings.

See the full set of tools we use inside of Athyna & Open Source CEO here.

HOW I CAN HELP 🥳

P.S. Want to work together?

1. Hiring global talent: If you’re hiring tech, business or ops talent and want to do it 80% less, check out my startup, Athyna. 🌏

2. See my tech stack: Find our suite of tools & resources for both this newsletter and Athyna here. 🧰 

3. Reach an audience of tech leaders: Advertise with us if you want to get in front of founders, investors and leaders in tech. 👀