EN
Back to the archive

The archive · Developer & Business Tools · Product decision · 2022–2025

Juicebox bets LLMs replace keyword recruiting; 2,500 customers, $30M from Sequoia

Two founders in YC S22 bet LLMs could find candidates keyword search misses; PeopleGPT reached 2,500+ customers and $10M ARR, then a $30M Sequoia Series A.

Juicebox

The betLLMs can find 'net new' candidates keyword search misses by reading profiles, websites and public pages like a human recruiter — AI search replaces manual sourcing.Scaling

What the business is

An AI recruiting search engine (PeopleGPT) that finds and ranks candidates from natural-language queries and then emails and schedules them.

How it started

During Y Combinator's Summer 2022 batch, David Paffenholz and Ishan Gupta built a small GPT-powered hiring tool. After two more years of iteration they launched PeopleGPT in late 2023.

What happened

PeopleGPT spread by word of mouth from small startups to companies like Cognition, Ramp and Perplexity. By 2025 it had 2,500+ customers and over $10M ARR, built by a four-person team with no salespeople.

How it ended up

Raised a $30M Series A led by Sequoia Capital, announced Sept 25, 2025, bringing total funding to $36M; still scaling in 2026.

Background

Juicebox is an AI recruiting search engine founded by David Paffenholz and Ishan Gupta during Y Combinator's Summer 2022 batch. Hiring software had used machine learning to match keywords across resumes and LinkedIn profiles for years, but a recruiter still had to open and judge each candidate by hand. The founders bet that LLMs could do that reading — understanding projects, skills and context the way a human recruiter does.

The result, PeopleGPT, launched in late 2023: recruiters type natural-language queries and the engine searches professional profiles, personal websites and other public information. Because it reasons about what a candidate built rather than matching terms, it surfaces 'net new' people that keyword searches miss. Once candidates are found, an agent emails them and schedules intro calls on the recruiter's behalf.

Adoption came without a sales team. Cognition, Ramp and Perplexity were early customers, and by September 2025 Juicebox was serving 2,500+ customers with over $10 million in annual recurring revenue — reached by a four-person core team. Sequoia partner David Cahn first heard of the company from a founder who had hired more than a dozen people without a professional recruiter; Sequoia led a $30M Series A announced on September 25, 2025, taking total funding to $36M.

What has to be true

  • Pairing a brand-new capability (LLM reasoning) with an old friction (manual vetting) sold itself, no sales org required.
  • Reading the whole public web, not just resumes, found exactly the candidates everyone else missed.
  • Self-serve word of mouth built 2,000 customers while the company still had four people.
  • Closing the loop mattered: finding candidates is useless unless someone contacts and schedules them.
  • Sequoia's recruiter testing the product internally gave the firm first-hand proof before the term sheet.

What can be applied

Do not fight incumbents on their own axis (keywords); move to the axis where a new capability is categorically better, and let delighted users double as your sales team.

Aftermath

Between the 2,000-customer milestone and the September 2025 announcement, Juicebox added eight employees beyond the founding four, still without a sales team. The Sequoia round was meant to scale inbound, with Sequoia arguing Juicebox could become a startup default like Stripe — the first thing founders use to hire. As of September 2026 the company continues to operate and compete with legacy talent platforms such as Eightfold, which are retrofitting AI search onto their products.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases