Company teardown · Listen Labs
For two years, Listen Labs was six people. Then came an eighteen-month sprint to ~$2 billion.
Two Harvard classmates built an AI interviewer in a hacker house and spent two years with almost nobody else in the room. When the product finally caught on, they hired 116 people, two in five of them to sell it, and Salesforce agreed to buy the company for a reported ~$2 billion. We read the LinkedIn profiles of all 122 people to see how the team was really built. This is the story the profiles tell.
0.1 — The whole story · start dates from 122 LinkedIn profiles
Six people for two years. Then 116 hires, and a buyer.
Everything here is public — LinkedIn, four founder podcast appearances, the Listen Labs blog, and press coverage of the billboard and the deal. The reading is ours.
§ 01 — Part one of four · Two years, six people
The real startup was six people in a room.
January 2023 to April 2025. A viral app, an AI interviewer built to understand it, and nearly two years the CEO later said did not look like a rocket ship.
1.1 — Who was in the room
Two founders, two engineers, and two people from Bain who did everything else.
— The CTO, on the Series A
“We were very small. I remember our deck — we were six people when the Series A.”
— And the roster
By the public launch in April 2025 the roster shows eight: a founding insights strategist joined in February and a founding GTM leader in March. Six is the team that built the product; eight is the team that launched it.
1.2 — What the six built
An interviewer built to understand their own app — and a panel that became the real product.
01 · First
02 · Then
03 · Finally
On the panel: “That's actually where we spend 80% of our engineering resources.”
— Alfred Wahlforss, January 2026
“For nearly two years, Listen Labs did not look like a rocket ship… we spent ~18 months in what felt like limbo.”
Then: “About 9 months ago, it finally clicked.”
— What Listen sells, in one sentence
A research study, end to end: find a thousand qualified people, have an AI interview each of them in their own language, and turn the transcripts into themes, quotes and charts in days. Sold mostly to marketing teams at big companies. Sweetgreen's six-week study took four days.
§ 02 — Part two of four · Build the engine
The order after the Series A: “Go build a go-to-market engine.”
2025. Six people become 27, and almost every hire is the first of its kind.
2.1 — A year of firsts · 2025
Twenty-one hires, in the order a company gets built.
2.2 — The billboard · September 2025
A $2,000 billboard with no design — and a puzzle that was secretly the product.
— Larkin Street, Nob Hill
https:// {64659, 123310, 75584, 8138, 38271}
— Translated
No logo, no company name. The numbers are tokens — the units a language model uses instead of words. Decode them and they spell a web address.
2.3 — What the billboard did
One hire. A year of press. And a million-dollar sequel.
It went viral on X within days. The SF Standard, VentureBeat and TechCrunch wrote it up, and a year later every article about the Salesforce deal retold it. The winner got his trip to Berlin. Florian's own accounting is modest: “We hired actually one person, not directly from the billboard. The main thing it helped us was to build a brand.”
A recruiting stunt that worked as marketing. Read it that way, and it was worth far more than $2,000.
Talent brief
Want the full visual breakdown?
Download the PDF version of this teardown — the headcount chart month by month, the six who built the product, the billboard costed, every quarter's hires by function, and five lessons for founders deciding when to start hiring.
§ 03 — Part three of four · The scale year
Ninety-five hires in nine months — sixteen of them engineers.
January to September 2026. The money arrives, the leadership layer arrives with it, and New York fills with people who sell.
3.1 — January 2026
The Series B was announced on the 14th. Twelve people started that month.
Ribbit Capital led a $69 million round at a $500 million valuation; Ribbit had been a customer first. The people who arrived with the money were the operating layer the company had never had.
3.2 — What each quarter hired · start dates from 122 profiles
Every quarter added three to five sellers and researchers for each engineer.
— Before 2026 · 27 people, January 2023 – December 2025
— January – March 2026 · 25 hires · Series B, January 14
— April – June 2026 · 33 hires · New York fills up
— July – September 2026 · 37 hires · agreement signed September 30
Engineering was a quarter of the company before 2026. Since then it has been between 12% and 22% of each quarter's hiring, while growth, sales and insights took roughly half to three-quarters. Three “engineer” titles — GTM, Growth and Marketing Engineer — are counted with the sellers, where the work is.
3.3 — Where the 95 came from
Hired from the industry it replaces — and from people who had sold AI to big companies before.
Simile, the synthetic-panel company we tore down in September, recruited from customer-experience incumbents instead. Nobody has moved between Listen and Simile, and nobody came from the AI-native peers — Outset, Strella, Conveo.
3.4 — Two cities, two companies
New York barely existed in March. By September it was 30 people — and one engineer.
The first New York hire started in January. The next 29 arrived between April and September, as the Soho office opened and filled. Madeline Mandell, who had moved the other way a year earlier, calls them “the Listen Lads”. Twenty-six of the thirty work in growth, sales, insights or operations; the one engineer joined from Datadog in July.
The engineers the billboard was for sit in one office. The people the price was for sit in the other.
3.5 — The engineers · 23 of 122
He said he needed a hundred engineers. He hired 23 — most of them one at a time.
— The CEO, on the pitch
“Luckily my co-founder is like the national champion in competitive programming in Germany.”
— What the roster supports
Third-party research says over 30% of engineering are IOI medallists. The roster supports two international medallists: the CTO and his former teammate, both from the founding era. Two more have national or regional olympiad records.
Twenty-three engineers is 19% of the company. Sixteen of them sit in San Francisco, within a few desks of the CTO. The CEO's public target was a hundred.
§ 04 — Part four of four · The deal
A signed term sheet, torn up for a buyer.
Summer to September 2026. Eighteen months after launch.
4.1 — How the deal came together
From $27 million to a ~$2 billion buyer in eighteen months.
— What Salesforce is buying
Agents that “uncover context behind customer and user feedback”, a panel of more than 50 million people and interviews in 120 languages — going into Marketing Cloud and Service Cloud.
— What Listen keeps
Under the agreement, Alfred stays CEO, Florian stays CTO, and “the same team will keep building the product you rely on, now inside Salesforce AI Labs.” “With Salesforce, we can bring Listen to every company in the world, much faster.”
Neither company disclosed terms; ~$2B is the reported figure. The deal is expected to close in Salesforce's fiscal fourth quarter, pending regulatory clearance.
4.2 — The team at signing
A company that had mostly not finished its first year.
The six who built the product are all still there, and under the agreement the founders keep their jobs. But the company Salesforce priced is overwhelmingly new. Four of every five people joined this year. The New York sales floor is at most six months old. What Salesforce is buying is a team that has barely met. The next twelve months are about keeping it.
5 — What we learned
Five things worth copying.
— The takeaway
Six people found the product.
A hundred and sixteen more made it a company —
and Salesforce agreed to buy it before most of them had finished a year.
Methodology & limitations
Every profile read — and every self-report labelled.
— Sources
- The LinkedIn company page and profile histories for all 122 people with a current Listen Labs role, collected October 4–5, 2026. Investors (7), advisors (3) and cards belonging to other companies are left out of every count.
- Four founder appearances — Sequoia Training Data, The Peel, Get the Check and Sequoia AI Ascent — and the Listen Labs blog.
- For the billboard and the deal: the SF Standard, VentureBeat, TechCrunch and SiliconANGLE, and employees' own LinkedIn posts for the later campaigns.
- IOI statistics, cphof and Codeforces for contest records; Google Scholar for the two publication records.
— Limitations
- 122 is what we read, out of about 125. Three anonymised cards are plausibly employees but cannot be read.
- No claims about degrees. LinkedIn withheld education pages after the first 14 profiles; job histories are complete for all 122.
- Founder quotes are self-report, and so are the revenue and interview figures. The billboard's cost is the CTO's $2,000; press reported about $5,000.
- Every join count carries survivorship bias. Departed staff never appear on a current roster, so early headcounts are floors.
- The function groups are our reading of titles, not an org chart Listen Labs publishes.
- The Salesforce agreement had not closed at collection. Teardowns like this are how our searches begin — this one's on us.