Juicebox is a serious company. It raised $80M from DST Global in March 2026 at an $850M valuation, counts 5,000 customers, and its PeopleGPT search runs over a database it says covers 800M+ profiles from 30+ sources, with 41 ATS integrations, 21 CRM integrations, sequences and autonomous sourcing agents. Recruiters leave it anyway, for four reasons that have little to do with the size of the database.
Every candidate is a stranger. Juicebox has no connection to your email, calendar or team, so it can't tell you that your engineering lead worked with a candidate for three years or that your investor sits on their board. The best hire in the database and a random match look the same to it. Its own pitch is reply-rate optimization on cold sequences, which is the tell: the product is built around the assumption that nobody knows anybody.
Specific queries get worse results. Plain-English search over a profile database works well for “senior backend engineers in Berlin” and degrades on “people who have run payments infrastructure at a company that went through a Series B”, because the database holds titles and employers, not what the person did.
It's recruiting-only now. PeopleGPT started as a general people search engine. Juicebox has since rebranded as an AI recruiting platform, PeopleGPT is the name of the search feature inside it, and the homepage says “win the talent war”. If you were using it to find customers, investors or partners, you are no longer the customer.
The seat price, and what sits on top of it. Starter is $119 per seat per month on monthly billing, or $99 per seat billed annually, with 500 email-and-phone credits. Growth is $199 a seat, or $179 annually, capped at five seats. Agents are another $199 per agent per month. The free plan is limited searches on one seat. For an agency with six recruiters the entry bill is over $700 a month before an agent is switched on.