He Sold His First Startup at 21. Then He Spent Three Years in Silence Building Airtable.

This is the story of three friends who met at Duke University, went their separate ways across Google, Microsoft, and Salesforce, and then came back together to build Airtable, one of the most quietly powerful no-code platforms on the planet.

Three Different Worlds, One Shared Frustration

The three met while at Duke University; a shared fascination with technology, video games and programming brought them together. All three tech geeks were a part of the same social scene, bound together by an instinct that the software available for the world just wasn't as good as it should have been.

After graduation, the three of them took off in entirely different directions, each taking a piece of the puzzle that would later form Airtable.

Liu grew up in College Station, Texas, the son of two immigrant parents from China who instilled in their child a work ethic they had never known before, as well as a sense of perfectionism. At Duke, Howie studied Mechanical Engineering & Public Policy. On the side, he learned programming, and upon graduating, chose to do instead the real work of actually building something. While still in his sophomore year, the twenty-year-old started Etacts, an application that automatically maintained a contacts list by reading email and calendar data. He sold Etacts to Salesforce less than one year later for one year at twenty-one.

After his acquisition, Howie worked as a product manager at Salesforce. As part of one of the largest software corporations on earth, he saw how organisations really used their software. Not the way the instructions implied, but how organisations actually used the software. This difference, he discovered, was a huge gap between the technical world and the non-technical; those with problems either had to wait months to receive their own version of a given software application or resort to using spreadsheets that were never really meant for the sort of complexity. This assumption in every piece of software was that computers were meant for technical users, while the rest just had to wait.

Andrew Ofstad grew up in rural Montana, where his childhood pastime was building treehouses with his two brothers using whatever materials were available. That lifelong skill was applied during his education at Duke, where he earned degrees in both Electrical Engineering and Economics. He landed one of the top jobs in the world for tech grads: an Associate Product Manager role at Google, after Marissa Mayer helped start that programme. Android was his first assignment; he gained experience shipping software to the masses quickly and efficiently, something many developers never achieve in their careers. He then worked on the Google Maps application, where he tackled what some believed was one of Google's most challenging and demanding product redesigns in company history: streamlining the product's interface without sacrificing its capabilities.

The question of how you make something incredibly complicated truly simple became the central theme of Ofstad's career. And it followed him all the way to Airtable.

Emmett Nicholas pursued a degree in Electrical Engineering & Computer Science at Duke before beginning his career as a software engineer at Microsoft. After Microsoft, he got the one job that would arguably define the scope of his understanding as much as any other in existence: founding engineer at Stack Overflow, the developer question-and-answer platform now being used by hundreds of millions, and from which Emmett learned not only about the importance of architecture but, arguably even more importantly, about its stability under the strain of that many users.

Three companies, three unique disciplines, and one common idea built over years of exposure to systems that nearly worked but never truly did.

The Idea That Brought Them Back Together

In 2012, Howie left Salesforce and called Andrew and Emmett. The concept was, by description, stunningly simple and, by building, unimaginably complex:

What if anyone could build a database without having to write a line of code? And what if it were a full, relational database, not a scaled-down spreadsheet or an application designed around some limited version of a database, but the whole thing, with relational fields, linked records and a variety of views all bundled in the form of a user interface indistinguishable from that of a spreadsheet yet as capable as real software?

The vision was not to make a spreadsheet more sophisticated. It was to shrink the time it took for a person to conceive of a problem and solve it, regardless of the person's aptitude for computer programming. Howie called it democratic software development. Andrew heard it as the same fundamental problem he had grappled with at Google for years: taking enormous power and packaging it into a simple application. Emmett recognised the engineering challenge he'd been building towards while answering questions on Stack Overflow: architecting infrastructure that could hold any data, no matter how big, and operate at any scale, for any group of people.

All three answered yes.

Three Years of Building Before Anyone Saw It

Airtable was founded in Jan 2013. Then the three founders did something almost unprecedented in the startup world: they didn't launch, they didn't release an MVP, they didn't build in public and get strangers to comment on a prototype. They went dark. And they built.

For three years.

Andrew managed the product and design. He had the same approach he brought to rebuilding Google Maps: starting with first principles and finding the simplest form for every component, then not being satisfied with a less complex version unless a more complex one truly earned its right. Each view that Airtable had-the grid, the kanban board, the calendar, and the gallery- was built so that it felt as if it had been around forever. The product's aim was an interface that someone who isn't technical could sit down with on Tuesday afternoon and be functional by Wednesday morning.

Emmett managed the architecture. Having every piece of data be connectable to any other piece of data, in any shape a user could imagine, was technically difficult to do quickly and scalably. Emmett knew infrastructure. Experience on Stack Overflow can instil a sense of gravitas that many young startups lack. Emmett built the Airtable backend for traffic that wasn't there yet, but that he knew would show up if the product was any good.

Howie managed the vision and the standards for work quality. Howie grew up having witnessed excellence. Anything less than that, he wasn't going to tolerate. The three years that the team spent in pre-launch weren't years spent dallying. They were three years of genuinely repeatedly asking themselves, "Is it good enough yet?"

In 2015, the team raised its seed and A round and went public.

First Milestone: The Unexpected Breadth of Who Needed This

The product that 3 years of quiet development produced ended up with some bizarre users: farmers leading livestock; attorneys tracing court cases; NPOs lacking all tech expertise to design complicated donor databases; multi-site scientists designing complicated experiments; filmmakers dealing with complicated production schedules; Cond Nast managing complicated editorial workflows.

The sheer scope of it was it. Andrew had designed it with intelligibility and stretch as key components, features that were necessary regardless of the near-infinite range of problems. Emmett designed it to support whatever data types those problems produced with underlying systems that were more than up to the task. Howie laid the groundwork that enabled non-technical users to create relational databases without even realising they were doing it.

The company was taking flight from seed to A to B, through Thrive Capital and Coatue Management, to the first $1.1B Series C in 2018, which officially made Airtable a unicorn.

Howie said something that reporters found stunning in 2018: "Everyone thinks what we're building is going to be an Excel or Google Sheets replacement; what we're going to do is build the next Microsoft or Apple." He didn't sound cocky-he sincerely thought that was the outcome of truly democratic software creation.

The Enterprise Pivot and the Growth It Unlocked

So, one of the key moves Airtable made is not obvious at first. In 2018, the company decided to begin serving enterprise clients alongside prosumer and SMB clients that had funded earlier growth. This move to the enterprise required building out security compliance, single sign-on, more granular permissioning, and the legal/procurement framework that large enterprises require before committing to an application. All of this did not change what Airtable was, but it did change what Airtable had to become around the edges. It changed what Airtable had to become to be used by large enterprise employees.

And Airtable has paid in staggering terms. Revenue for the enterprise line was 100% YoY growth for 2023, and net dollar retention, the metric of dollars in existing accounts growth, was 170% - ahead of Asana (130%) and Monday.com (120%). Existing customers did not just renew; they expanded the application's use across their enterprise and increasingly viewed it not just as a team tool but as mission-critical. Clients like Netflix, TIME, JetBlue, Shopify and over 80% of the Fortune 100 used the application. In 2023, Airtable generated $375m of ARR, 50% YoY growth, and Airtable expects to generate $478m in ARR in 2024.

The Hard Years: Layoffs, Valuation Reset, and the Path Forward

Last year, it raised $735 million in Series F financing at a $11.7 billion valuation, at a peak time for venture financing. In total, it raised $1.4 billion, but the market broke. Increased rates and a reckoning across software valuations sent it plummeting. In 2024, estimates place the company's secondary valuation at $4 billion, around a third of the 2021 high. Last September, it laid off 237 employees, around a quarter of its staff, as it leaned into enterprise and profitable growth, which was devastating for those involved. The foundations of the business never broke. ARR rose, enterprise retention stayed solid, cash-flow broke even at the end of 2024, whilst the business grew at ~30% per year. The team that survived this period is lean, effective and appropriately sized for future growth, unlike at the peak of venture financing.

The Refounding: Going AI-Native

Howie declared in June of 2025 that Airtable needed to be "refounded as an AI-native platform and not simply updated". Refounded. The product architecture was rebuilt: Omni is the front-door conversational AI interface that accepts any natural-language description of a workflow and creates a production-ready Airtable base and interface. Field Agents is the execution layer of AI that performs research, content enrichment, and routing for any Airtable application at scale. In October 2025, they acquired DeepSky, an AI superagent startup, and they hired their CTO from OpenAI.

This market already existed, as was evident from Cobuilder, which came out in July 2024. Cobuilder became the most adopted feature of any Airtable launch to date and allows you to build an application without any coding by simply telling Airtable what you want it to do in natural language.

For 3 years, the three founders had been working quietly to develop a product that would make software development accessible to anyone. Today, after twelve years, they were building that same product around the most powerful tool in the history of humankind, for the same reason they'd built it in the first place.

Where Airtable Stands in 2026

Today, approximately 500,000 organisations, 166,000 of whom pay for the service, and more than 80% of whom are part of the Fortune 100, use Airtable. Airtable brings in approximately $478 million in annual recurring revenue and has about 947 employees. The company might even IPO: CFO Ambereen Toubassy said, "It is a consideration," and that they had not filed as of December 2025.

Howie Liu continues to be the CEO and is just as careful. Andrew Ofstad continues to be CPO and asks the question he has always asked: "How do you make something very powerful look simple?" Emmett Nicholas continues to serve as CTO and is writing infrastructure that needs to carry more than it currently does.

Three friends who went their separate ways while studying at Duke, then came back together when the precise set of talents needed to build the service materialised, building one of the most widely used productivity platforms on earth for 12 years.

And they are still building.