Thomas Chan's first recruiting startup taught him that enterprise HR departments don't buy point solutions anymore. They want systems that do everything, or at least pretend to.
His second attempt, Lantern AI, crossed $1 million in annual recurring revenue this past October, about 11 months after the San Francisco-based company started selling to enterprises, according to self-reported figures on its Y Combinator profile and a LinkedIn update Chan posted that same week. The client list reads like a corporate sampler platter: Crypto.com, Hyatt, law firm Baker McKenzie, Hong Kong construction giant Gammon, and three universities across the city's career services offices.
What Lantern sells is what every AI recruiting startup now claims to sell, which is the whole workflow. Candidate sourcing, resume screening, live AI interviews in 40-plus languages, debrief note capture, analytics dashboards. The pitch is that hiring managers define what "good" looks like in a 15-minute calibration session, then Lantern's software handles everything up to the shortlist.
Whether that actually works at scale, or whether enterprises are buying because they feel pressure to deploy AI somewhere and recruiting seems safer than, say, legal review, remains an open question. But Chan and his co-founder Justin Tang have at least convinced some sizable organizations to pay for it.
The product started narrower, then customers asked for more
Chan and Tang didn't launch with an end-to-end platform. In mid-2024, Hong Kong media covered what was then called "Lantern AI Interview Assistant," a more focused tool helping local employers screen candidates at volume. Chan told HK01, a local news outlet, that early enterprise clients including Gammon Construction and property developer Sino Group saved 800 hours processing 750 candidates and cut hiring time by 68 percent using the AI interviewer, according to company-provided metrics reported in May 2024.
Chan had been here before, sort of. His previous venture, Career Hackers, was a Hong Kong recruiting startup that placed in the top 20 of HKTDC's Start-up Express competition in 2025 and counted Baker McKenzie and L'Oréal among clients. That experience gave him a front-row view of how enterprise hiring actually works, which is to say messily and with lots of spreadsheets.
"We started as an AI interviewer," Chan wrote in an October LinkedIn post announcing the Y Combinator acceptance. "Then our clients started asking harder questions."
The company rebranded from "ourlanterns.com" to "lantern.md" and rebuilt the product into what it now calls an "agentic recruiting platform." The shift happened before the founders joined Y Combinator's recent cohort, which ran from October through December. Exactly when the pivot occurred isn't clear from public materials, but the timeline suggests it was sometime in the middle of last year as enterprise customers kept requesting features beyond just interviews.
How Lantern says it works
The platform revolves around two proprietary data models, at least according to Lantern's marketing. The "World People Model" claims to index one billion professional profiles and powers candidate sourcing that goes beyond keyword searches. Recruiters can query by what someone built, career trajectory, or the context of companies they worked for, per the product site. The "Company Hiring Model" learns from a team's past decisions, interview feedback, and applicant tracking system history to calibrate what Lantern calls a hiring "taste."
Taste is a deliberately soft word for what is essentially pattern recognition trained on historical hiring data, which raises all the usual questions about bias that enterprises are supposed to care about but often don't audit closely.
The workflow itself starts with that 15-minute role calibration session. From there, Lantern sources candidates, reviews applications against the bar the hiring manager set, conducts adaptive interviews that adjust follow-up questions based on candidate answers, captures meeting notes from human debrief calls, and surfaces analytics by connecting to a company's existing ATS. The startup claims integrations with more than 30 applicant tracking systems including Ashby, Lever, Greenhouse, Workday, and SAP SuccessFactors, per an October LinkedIn update.
The AI interviews run live and produce transcript-backed assessments. A client quote on the product site, attributed to Hyatt but without a named spokesperson, said the software "ran the first round at scale without losing the warmth candidates expect from our brand." That's the kind of testimonial that sounds good in a sales deck but leaves a reporter wondering what "warmth" means when delivered by an algorithm.
Chan studied law at Tsinghua University before getting into recruiting, which perhaps explains the emphasis on structured evaluation criteria. Tang, listed as CTO on the Y Combinator profile, led technical development. The Y Combinator directory lists four team members, though LinkedIn lists the company size as 11 to 50 employees. That discrepancy likely reflects contractors, advisors, or optimistic categorization for recruiting purposes.
Clients span continents but most traction looks Asia-focused
Lantern's customer logos span hospitality, crypto, construction, real estate, and professional services. Hyatt and Grand Hyatt. Crypto.com. Gammon. Sino Group and Nan Fung Group, both major Hong Kong property developers. Baker McKenzie and its regional affiliate Wong & Partners.
Three Hong Kong universities use the platform: City University of Hong Kong, Hong Kong Polytechnic University, and Lingnan University. CityU's career services office published a user manual for students practicing interviews with Lantern's AI, which suggests the universities are at least somewhat committed deployments rather than one-off pilots, though whether they're paying customers isn't specified.
The startup says it serves clients across 10 markets, per its Y Combinator profile, but hasn't disclosed revenue breakdown by geography. Most visible traction sits in Asia-Pacific. The Hong Kong Science & Technology Parks and several local conglomerates appear among logo clients. Still, U.S. incorporation (Lantern AI Inc., 2261 Market Street, San Francisco) and Hyatt's participation suggest some North American adoption.
A Baker McKenzie client quote on the site, also unattributed to a named individual, praised "a structured, consistent way to assess every candidate against the same criteria." Law firms love that language. Standardized evaluation reduces liability and sounds defensible if anyone questions hiring outcomes later.
The market is extremely crowded
Lantern is hardly alone in trying to automate recruiting. HireVue has been offering voice-based AI interviews for years and counts a long list of Fortune 500 clients. Eightfold AI runs a Talent Intelligence Platform with deep-learning models trained on external labor market data and has raised hundreds of millions in venture funding. Paradox's Olivia chatbot focuses on high-volume hourly hiring for franchises and retail chains. Metaview, another startup, markets itself as an "agentic recruiting platform" handling sourcing and screening, which is nearly identical positioning to Lantern's headline.
Chan and Tang's bet is that enterprises want a single system spanning the full pre-hire workflow, not five different tools duct-taped together with Zapier integrations. Whether that's true or whether HR departments will keep buying best-of-breed point solutions depends on who you ask. Enterprise software buying is often less about optimal architecture and more about which vendor got to the VP of Talent first.
The company emphasizes compliance readiness. It claims ISO 27001 certification, SOC 2 Type II certification, and GDPR compliance on its homepage, though the underlying audit reports aren't publicly posted. Enterprise IT requirements like single sign-on, SAML, and SCIM appear in product documentation, which matters when you're trying to get past IT security review at a multinational law firm or hotel chain.
Lantern also launched something called Lantern People, a connector that lets AI agents like Claude and others access people and company data for recruiting research. The product page lists a free trial with three days of unlimited search, 200 contact reveals, and $5 in research credits. It's a developer-focused motion that mirrors broader SaaS trends of embedding into agent workflows, though it's unclear how much traction that side product has versus the main enterprise platform.
No announced funding beyond Y Combinator
Lantern hit $1 million ARR without any disclosed venture rounds beyond Y Combinator's standard investment, which is typically $500,000 for seven percent equity in recent batches. The company updated its privacy policy on October 9 and terms of service on September 26, signaling active commercial traction or at least legal housekeeping ahead of larger deals.
Chan wrote in October that founder-led LinkedIn content drove "50 to 80 inbound leads per week" after Y Combinator, though that figure is self-reported marketing commentary, not audited pipeline data. Take it with the usual grain of salt that attaches to founder social media posts.
The startup runs an enterprise sales motion with "book a demo" calls to action. No public pricing appears on the site, which is standard for enterprise SaaS that prices based on employee count or usage volume. Y Combinator's backing brings credibility and a network of potential customers, but Lantern hasn't disclosed whether it's raising a seed round or operating on early revenue and YC capital alone.
Chan appeared on the Fondo START podcast in September and offered a glimpse of product philosophy. The system learns hiring preferences from a team's decisions, challenges unrealistic job requirements, and pulls from more than 50 data providers to build candidate profiles. The company's tagline is "Turn Your Taste Into Hires," which underscores the positioning: Lantern doesn't replace hiring judgment; it encodes it.
What the numbers mean and what they don't
Eleven months from founding to seven figures in ARR places Lantern among faster-scaling B2B SaaS companies, particularly for a product selling into notoriously cautious enterprise HR buyers. Whether that pace continues depends on factors the startup hasn't disclosed. Gross retention, expansion revenue, sales cycle length, contract concentration. How much of the $1 million comes from a handful of large contracts versus distributed smaller deals matters a lot for projecting future growth.
The client roster suggests at least some large enterprise adoption. Crypto.com and Hyatt aren't small pilot buyers. But the company declined to specify contract sizes or deployment scale. Those three universities using the platform for student interview practice may or may not represent paid contracts. Career services offices often pilot tools for free in exchange for student access, and universities love being able to tell students they're learning with cutting-edge AI.
Competitors with years of head start and venture war chests will test Lantern's ability to differentiate on product depth rather than just speed to market. Chan and Tang's prior startup gave them domain expertise and early customer relationships. Baker McKenzie appeared on both Career Hackers' and Lantern's client lists, which suggests the founders are levering existing relationships. But translating a Hong Kong-grown product into a global enterprise platform is a different challenge than winning regional adoption.
The founders haven't announced next milestones publicly. For now, the story is 11 months, $1 million ARR, a Y Combinator badge, and a bet that enterprises want one AI system to run recruiting end-to-end. Whether that bet pays off likely depends less on the technology and more on whether HR departments decide integrated platforms are worth the switching cost from whatever patchwork they're already using.
