Before Austin Hughes built software for finding the right buyer at the right moment, he was a college student trying to find one person in New York. Hughes was at Rice University in Houston, studying Mathematical Economic Analysis. His classmates were largely looking nearby. He wanted a job in a city where he had no network, so he searched online for a Rice alumnus whose career resembled the one he imagined. Then he wrote a careful email. The alumnus replied before Hughes had much time to worry about it. Later, the job came through.
It is a neat origin story because it contains the entire argument of Unify in miniature. A stranger can welcome a message when the sender has chosen well, arrived at a useful moment, and done enough work to make the note relevant. The difference between an interruption and an opportunity is often context.
Hughes began his career in 2016 as an investment banking analyst at Centerview Partners. He moved west in 2019 to invest at SoftBank Vision Fund, then joined Ramp in 2020 as employee 45. The sequence looks like a steady march from spreadsheets toward software. It was also an education in how companies are valued, funded, measured, and finally operated from the inside.
The side projects were the straight line
The résumé hides the messier preparation. In his early twenties, Hughes spent nights and weekends making things. Simon's Wealth was a blog about real-estate investing. Fisheye Surf was a marketplace where surfers could find photographers. He considered a direct-to-consumer shoe company and, in his own later telling, was glad he left it at the idea stage. None became his career. Each offered a small rehearsal for it.
Fisheye Surf mattered in another way: it introduced him to designer Hyewon Kim, who would later join Unify. The project itself had been built on Bubble, a reminder that the earliest useful question is rarely whether the technology is impressive. It is whether somebody will use the thing. Hughes came to describe those experiments as his education in going from zero to one.
Write the blog. Mock up the marketplace. Test the idea before polishing the machine. Side projects gave Hughes practice crossing the awkward distance between an empty page and a real user.
At Ramp, that instinct acquired an operating system. The growth function behaved more like a product team than a traditional marketing department. It ran in sprints, wrote specifications, tracked metrics, and treated programs as bets. Hughes helped grow the group from one person to more than 25. At one point, it was running more than 100 experiments a quarter.
“The thing that matters most is the impact rather than the actual solution that you're building.”Austin Hughes
The practical trick was scope. A proposal that looked like two weeks of engineering could sometimes be reduced to an hour, a manually assembled CSV, or a narrow test. Hughes divided work into MVPs and full builds. The MVP earned information. The full build earned scale only after the information was promising. That habit sounds obvious until a team falls in love with an elegant system before checking whether the system changes an outcome.
The Hughes experiment ladder
When the workflow becomes the company
Outbound sales was one of Hughes's focus areas at Ramp. He saw representatives spending time on research, enrichment, CRM maintenance, and sequences while important buying signals sat across disconnected systems. In mid-2022, before ChatGPT arrived, he began building with GPT-3. The model's generative ability was interesting. Its potential to make a chain of decisions across those manual workflows was more consequential.
Hughes left Ramp at the start of 2023 and built Unify with Connor Heggie, a machine-learning engineer whose background included Scale AI and Helm.ai. They spent a year in the quieter work before launching publicly in January 2024. Unify's early label was “warm outbound”: find evidence that an account may need what a company sells, identify the relevant people, and use that timing to begin a better conversation.
The house analogy is Hughes's cleanest explanation. A seller would not stop a random person on the street and expect to sell them a home. Knowing that the person is already looking in a particular neighborhood changes the conversation. Unify applies that principle to signals such as website activity, a champion changing jobs, a company hiring for a revealing role, product usage, or repeat engagement with an email.
The software ingests data from CRMs and other sources, maps companies, people, and activities, then runs “plays” when conditions match. A play may qualify an account, find the right contact, research the context, write a message, add a person to a sequence, or alert a human. The more interesting promise is not that AI can send more email. Any system can increase volume. The promise is that software can be choosier.
Distribution goes on the calendar
Unify's public launch provided its own lesson in signals. The seed announcement generated roughly 420 demo requests within 48 hours, most of them attributed to LinkedIn. Hughes had not been a regular LinkedIn writer. By spring 2024, he was posting five times a week. He now keeps a note on his phone for ideas that arrive while walking or hiking, then reserves four or five hours on Sunday to write the week's posts.
He writes the final prose himself. In a 2025 interview about his process, Hughes said AI did not yet produce the style he wanted and that the words mattered too much to delegate. It is a wonderfully inconvenient position for the chief executive of an AI company: automate research and repetitive work, but keep the judgment-heavy last mile in human hands.
The company's funding announcements became product launches for distribution. Unify paired press with product releases, customer material, a refreshed website, coordinated employee posts, outbound plays, and, during the Series B campaign, personalized Cameo videos from B-list celebrities. The joke had a spreadsheet behind it. Hughes later reported that the July campaign helped produce 885 meetings and $25 million in new pipeline during the month.
There is a sharper insight here than “founders should post.” Hughes treats attention as another product surface. An audience sees the company repeatedly, learns its language, and may eventually encounter it at the moment a problem becomes urgent. Content is not separate from the sales system. It creates signals for it.
“Distribution is the advantage that we're looking for at a startup.”Austin Hughes
From systems of record to systems of action
The company changed as it grew. Unify raised $6.6 million from OpenAI Startup Fund, Thrive, and Emergence around its public launch, followed by a $12 million Series A in October 2024. Nine months later, Battery Ventures led a $40 million Series B. Unify said revenue had grown eightfold over the preceding year and named Cursor, Perplexity, Flock Safety, Airwallex, and Together AI among its customers.
Hughes described the investor questions as a progression. At seed, the bet was on the founders, before product or customers. At Series A, investors wanted evidence that customer acquisition could repeat. At Series B, retention, expansion, unit economics, competitive position, and a route toward a much larger revenue business became central. The questions changed because the company had earned new ones.
His management changed too. At Ramp, Hughes had used a situational approach influenced by Andrew Grove: offer close tactical direction when a person or task needs it, then delegate as familiarity grows. A founder's constant context switching made that harder. Unify began emphasizing autonomy and what the team calls “slope over intercept” - the rate at which someone can learn and expand, not only the experience visible on day one.
In 2026, Hughes summarized two lessons from Unify's reported 24-fold growth in 2024. First, many business problems eventually reveal themselves as people or alignment problems. Second, a company inventing a new category has to build the explanation as deliberately as it builds the product. Unify's language moved from autonomous SDRs to warm outbound, then toward a broader “system of action” for revenue. Each phrase was an attempt to make an unfamiliar workflow legible.
The useful restraint
Automation is an easy story to tell as replacement. Hughes has sometimes spoken bluntly about low-value sales development work disappearing while better-paid, more capable roles rise. Yet his product view retains a boundary. He has argued that AI inbox agents remain risky when they lack the context to distinguish a disposable pitch from a consequential message. A false positive can erase an opportunity. Human review still matters.
That boundary clarifies the company he is trying to build. The point is to remove repetitive work while bringing a seller into the loop when the seller's judgment and capacity for connection matter. Growth engineers can construct the workflows. Agents can research, enrich, and prepare. People can handle the moment that should feel like a conversation rather than a sequence step.
Away from work, Hughes has written about photography, music, skiing, surfing, and investing. His personal site still carries images and playlists from the period before Unify. They make the résumé feel less inevitable. The founder did not emerge from a perfectly optimized plan. He followed a chain of small experiments, moved between New York and San Francisco, learned from operators at Ramp, and kept notes when ideas arrived.
The same pattern now runs through Unify: notice something, reduce it to a test, watch what happens, and build only after the world answers. AI makes the machinery faster. Hughes's contribution is a quieter discipline about where to point it. The best growth system is not merely busy. It learns.