David Paffenholz’s case for starting over

The recruiting marketplace had failed. David Paffenholz thought his startup journey might have ended with it. A week later, he and his co-founder, Ishan Gupta, returned to a question small enough to ask without a pitch deck: did they still believe they could build something big? They did. They stayed close to recruiters, listened to the complaints, and kept working. The marketplace had run out of road. Its most useful lesson was still available.

That is a revealing place to begin with Paffenholz, the German co-founder and CEO of Juicebox. Today his company sells software for finding and contacting candidates. Its history is full of decisions to stop doing something: abandon a product, leave a city, narrow a problem. Read backward from a funding announcement, each decision can look inevitable. At the time, it required giving up a perfectly serviceable explanation of what the founders were doing.

“Do we still believe we can build something big?”

The question Paffenholz recalls asking with Ishan Gupta

A growth education, then the exit

Paffenholz grew up in Germany and came to the United States for college. At Harvard, he studied economics, completed his degree in three academic years, and served as a teaching fellow in economics. His LinkedIn profile also records work organizing the German American Conference at Harvard. Before there was a recruiting company, there was already a pattern of bringing people together.

His early work included positions at New Enterprise Associates and Moonfare, as well as international growth at Snap. The Snap stint, from mid-2020 to mid-2021, began as a three-month internship and stretched to almost a year. He worked on expanding usage in international markets, seeing how an established consumer product measured whether people actually wanted to return.

That experience supplied a demanding comparison for his own experiments. He and Gupta built a music discovery app that attracted roughly 50,000 users. People shared it on TikTok. The audience was real; the habit was weaker than he wanted. They shut it down. Fifty thousand users would make a handsome slide in a presentation. It made a less convincing reason to spend the next several years.

The recruiter’s chair was the classroom

Juicebox dates its founding to 2022 and went through Y Combinator’s Summer 2022 batch. Paffenholz and Gupta tried a recruiting marketplace, taking on sourcing and bringing talent onto the platform themselves. The effort gave them an intimate view of work they had hoped to improve. A marketplace can look elegant from a distance. Up close, somebody still has to find the people.

The founders’ education continued with recruiters and talent acquisition leaders. They spent time watching the work, learning where the hours went and which frustrations recurred. Paffenholz’s earlier exposure to NEA’s talent team had shown him both internal recruiting and the hiring problems of portfolio companies. Now the question was whether those observations could support a product people would keep using.

He came to focus on the gap between a candidate matching a set of filters and a candidate matching the actual intention behind a role. A title and a location could narrow a list. They could leave the recruiter with a great deal of reading. The task was to make more of that judgment available inside the search itself.

Plain English meets a very large list

PeopleGPT, the search product associated with Juicebox, let recruiters describe whom they wanted in ordinary language. Instead of making a person translate every hiring requirement into a complicated query, the software would help interpret it. The appeal was easy to demonstrate. A recruiter could recognize the question before knowing anything about the machinery answering it.

By May 2025, Paffenholz was describing a two-stage search across about 800 million public profiles. An initial pass narrowed the field using information such as titles, locations and skills. A second stage assessed candidates against more specific criteria supplied by the recruiter. Someone who had managed five account executives at a business software company represented a more useful requirement than a title alone.

The distinction matters to his story because it keeps the problem concrete. A recruiter has a role to fill, a limited day and a reason to distrust a promising list until the details survive inspection. Paffenholz’s product sits in that interval between finding a possible name and deciding the person deserves a conversation. The search bar is the visible part; the evaluation behind it carries the burden.

A launch is a crowded doorway

The early PeopleGPT launch brought the sort of attention founders hope for and servers tend to resent. Paffenholz has recalled expecting to be happy if 200 people tried the product. Around 30,000 tried it over the launch weekend. The team stayed up keeping the platform running as things broke. Interest had arrived faster than the product was prepared to receive it.

He later became careful about distinguishing a message that makes people sign up from a product that makes them stay. That distinction gives his earlier music app a second place in the story. A crowd at the entrance is an encouraging sight. The more consequential count comes after people have had time to leave.

By September 2025, Juicebox had reached a different set of milestones: $10 million in annual recurring revenue and 2,500 customers as Sequoia joined its investors. The company had earlier reached $1 million in annual recurring revenue with four people. Those figures describe a business earning repeat payments, a more substantial test than a launch video enjoying a good weekend.

The couch before the company office

There is a second sofa in this story, considerably less ceremonial than the one in an investor photograph. Before moving to San Francisco, Paffenholz flew there weekly for two months to attend founder meetups through Ali Rowghani’s program. He stayed on his friend Pedro’s couch. Proximity had a practical form: a place to sleep and another group of people to talk to.

In January 2024, he and Gupta decided to move from New York. Paffenholz has described New York as the period when the company was essentially the two founders iterating. San Francisco was where they began building a team. He liked the surrounding concentration of founders and the sense of technological change visible in the city’s autonomous vehicles.

His enthusiasm for building in America is also personal. He has written appreciatively about growing up in Germany while explaining why he wanted to build his company in the United States. The attraction, in his account, includes a culture receptive to underdogs and investors willing to back risky private businesses. For him, choosing a place to live became part of choosing how to work.

The software needs to show its working

Paffenholz’s public writing introduces a useful complication to the sales pitch for automation. He argues that recruiting systems should explain why they recommend someone and indicate when the available information is insufficient. A candidate with a smaller online footprint can be harder for software to represent. More searchable information does not automatically mean a fairer view of every person.

He has also warned that explanations generated by a model can sound plausible while being wrong. The recruiter retains responsibility for evaluating the result. That is a significant qualification from someone whose company benefits when customers entrust more work to software. It places the quality of the evidence beside the speed of producing a shortlist.

In another essay, he argues that automation magnifies the quality of the hiring process beneath it. Clear role definitions, deliberate sourcing and consistent evaluation give a tool something worthwhile to accelerate. Vague requirements get processed too. His argument asks customers to do some unglamorous preparation before enjoying the glamorous demonstration. Even the cleverest search cannot settle an argument a hiring team has not yet had.

A search company learns to hire

As Juicebox expanded, Paffenholz’s own organization became a test of his ideas about people. In September 2026, he described a company of roughly 90 employees, with no engineering managers and a five-day office policy. He said it had lost no employees in the preceding two years. That is his account of a particular period, rather than a promise about the next one.

He connected retention to independence, visible impact and meaningful equity. He also discussed a decision he had reversed: resisting an MCP server before changing his mind. The detail is useful because it lets the CEO appear as a participant in the learning process. Building software for other people’s judgment calls gives him no exemption from making his own.

The ambition remains tightly framed around sourcing. “We want to stay narrow on sourcing,” he has said. In an industry fond of expanding every product description until it occupies an entire department, choosing a boundary is revealing. Paffenholz wants more of the work of finding candidates to happen through Juicebox. A growing company still needs a clear answer about which work deserves its attention.

London, Fetcher, and the next search

On March 10, 2026, Juicebox announced an $80 million Series B led by DST Global, valuing the company at $850 million and bringing disclosed funding to $116 million. It said more than 5,000 recruiting teams used the platform. The money was earmarked for product development, international expansion and hiring. Those plans soon acquired addresses and counterparties.

On September 30, Juicebox announced its first international office in London’s Soho, serving a region where it already had more than 1,000 customers. It planned to grow the London team to more than 50 people over the following year. On October 6, Paffenholz announced an agreement to acquire Fetcher’s assets as that recruiting platform prepared to retire, with support for customers moving across.

The smaller details survive alongside the larger numbers. Paffenholz’s first paid work was tutoring in ninth grade. His repeat listening includes the German musician Peter Fox. In San Francisco, he has a weakness for Salt & Straw ice cream. The company has moved a long way from two founders wondering whether to continue. Its CEO still has to persuade people to choose something, return to it, and tell somebody else.