Lantern AI runs recruiting agents that sourced for Crypto.com, Hyatt and Baker McKenzie

A recruiter's taste for who fits a team sits in their head and their browser tabs. When the recruiter leaves, the taste leaves too. Hiring tools like Workday, SuccessFactors and Ashby record what already happened, the launch page says. They do not know who is open to a new job today, or what makes someone right for this team. Lantern AI launched on 28 October to answer both.

The company says it has processed more than 2 million candidates. Named customers on the launch page include Crypto.com, Hyatt, Baker McKenzie and Eight Sleep. The YC launch had nine votes in its first hours. That is the highest of any launch read on 29 October.

What the agents do

Lantern says it runs on two internal models, both its own. The first is a live map across what Lantern calls a billion professionals. It tracks role changes, product launches, team moves and who is hiring. The second is a hiring taste model, built from a company's interview notes and offer history. Every interview and every offer updates the second one.

Five steps make up the loop, the launch page says. The agent joins a kickoff call between the hiring manager and the recruiter. It checks the brief against who the company has hired and rejected before. It then sources from the people model, builds a shortlist and drafts outreach for the recruiter to approve. Applications and replies trigger an AI screening interview. Debriefs feed back into the taste model, so the next search starts smarter than the last.

The product runs on top of an existing applicant tracking system, or can replace it, Lantern says. Sourcing, review, AI interviews, meeting intelligence, scheduling and analytics can all be driven from Slack or an MCP server.

What a buyer should ask

The company's own claim is the 2 million-candidate figure, and the four named customers back it. What the page does not back is the match quality of those shortlists. It also does not say how often an AI screening gets the same answer as a human screening. A hiring team can run its own test. Have Lantern source and screen 50 roles the team has already filled. Count how many of the hires show up in Lantern's top three. That number tells the team what the taste model is worth on their own hiring.

Who built it

The two founders, Thomas and Justin, co-founded two earlier recruiting startups together, the launch page says. Thomas worked at Morgan Stanley and HSBC. Justin won three Olympiad gold medals in Informatics.

Source

• Primary source: Lantern AI on Y Combinator.
• Company site: Lantern.