Shensi
Ding

The scene: a Sweetgreen somewhere in San Francisco, 2019. Gil Feig keeps showing up to their regular dinners looking wrecked. Not hungover. Not overworked in the general sense. Wrecked by a single, specific, infuriating thing: integrations.

Gil was a founding engineer at Untapped, a recruiting startup. Shensi Ding was Chief of Staff at Expanse, a cybersecurity company that would later sell to Palo Alto Networks for around a billion dollars. Both companies were drowning in the same invisible tax: every enterprise customer expected your product to connect with their existing tools. HRIS systems, ATS platforms, CRMs, ticketing software. And building those connections, each one, was slow, expensive, and the work never actually ended.

Shensi and Gil had known each other since freshman orientation at Columbia. They'd been in the same CS classes, the same group projects, the same social circles. By senior year they were running the engineering student council - she was President, he was VP. After graduation they'd each gone separate ways: she toward finance (Credit Suisse, then Silver Lake), then back toward tech operations; he toward engineering. The Sweetgreen dinners were how they stayed in each other's orbits.

The integration problem wasn't new. But something about the combination of both of them hitting the same wall at the same time, at different companies, in different roles, made it feel structural rather than incidental. Not a bug. A feature of the SaaS market itself.

What followed was six to eight months of methodical research. They interviewed more than a hundred companies - not pitching anything, not testing product ideas, just listening. They talked to sales reps, partnership leads, product managers, engineers, customer success teams. They wanted to understand the shape of the problem from every angle before they even thought about solutions.

When pricing conversations came up organically - they didn't raise it, customers did - numbers like $50,000 to $300,000 appeared. That wasn't a price point test. It was a market signal. This problem was expensive enough that people would pay serious money to make it go away.

In June 2020, they left their jobs. They started building out of Gil's San Francisco apartment.

The insight that became Merge is simple enough to explain in a sentence: instead of every B2B software company building and maintaining dozens of separate integrations, they integrate once with Merge's unified API, and Merge handles the rest.

Simple to say. Genuinely difficult to build. The integration space is a graveyard of failed attempts - tools that covered five platforms, or ten, but couldn't achieve the breadth that enterprise buyers required. Shensi and Gil weren't trying to build a better integration tool. They were trying to make integrations into infrastructure - the kind you stop thinking about, the way you stop thinking about payment processing once you've wired up Stripe.

They launched in stealth for nearly a year, and when they went public in April 2021, they did something deliberate: they launched two integration categories simultaneously (HRIS and ATS, two of the most painful categories for HR tech and recruiting software). Not one. Two. The message to the market was explicit - this is a platform play, not a feature.

The self-serve model with a free tier removed friction. But both founders spent the next four to five months personally running back-to-back sales meetings globally, treating each one as an A/B test. What messaging landed? What demo structure worked? Who in the buying organization actually had authority? They were the first to find out that meeting with individual engineers was a dead end - decision-makers were heads of product and engineering. Every conversation taught them something.

"Founder-led selling creates a tighter feedback loop that makes your product better," Shensi has said. It also turned out to be the best training for investor pitches. The same story that closed enterprise deals resonated in term sheet conversations.

The AI wave turbocharged what was already working. As LLMs began proliferating in 2022 and 2023, AI-native startups discovered that their models were only as useful as the data they could access - and that data lived inside third-party systems: Salesforce, Workday, BambooHR, Jira, GitHub. Merge's infrastructure became critical plumbing for the entire AI ecosystem. Frontier LLMs and Fortune 500s alike began using Merge not just to offer integrations to customers, but to give AI agents the real-time data access they needed to act.

On co-founding

Most founders talk about co-founders in terms of complementary skills: one technical, one commercial; one builder, one seller. Shensi talks about Gil Feig as a life choice. "My co-founder will be the man of honor at my wedding," she's said. "Who you're going to be working with is a huge decision." They've known each other for over a decade. The trust isn't built - it pre-existed the company.

At Merge, the division of labor reflects that comfort level. Gil leads product and engineering. Shensi took on sales, commercial strategy, and investor relations. Neither role was assigned - they evolved naturally from where each person's energy was strongest.

On customer obsession as a survival strategy

The first Merge customer, Shensi has described, was like "a tough parental figure who's just disappointed in you all of the time." Enterprise buyers in the integration space had been burned before - by vendors who promised coverage and delivered partial solutions. Merge's early bet was to over-invest in trust: enterprise-grade security certifications, reliability, and support quality that wasn't strictly necessary for a company at their stage. It was necessary for the customers they wanted.

On fundraising as a learnable skill

The five years in investment banking gave Shensi a front-row view of how capital flows - who makes decisions, how they frame risk, what moves a deal forward. When she sat across the table as a founder, she was less surprised by the process. "There is a process and there is a game, and you need to learn it," she's said. Passion alone doesn't close a round. Understanding the mechanics does.

On the personal cost

She's unusually candid about what it takes out of you. "I don't know if I could do this again - sometimes the first one takes a lot of energy out of you." It's not a complaint. It's an acknowledgment that the "accidental confidence" she credits as a first-time founder advantage - the naivety that lets you attempt things you'd never attempt if you fully understood the difficulty - doesn't refill on demand. You spend it once.