The startup teaching AI to hire humans

The resume is a bad piece of technology. It flattens a decade of work into a column of job titles and dates, and it says almost nothing about the thing an employer or a hedge fund actually wants: what you can do that nobody else in the room can. Ethos, a London company founded in 2024, is betting that an AI voice call can find that thing - and that finding it is worth a lot of money.

The pitch cuts against the prevailing mood in artificial intelligence. Where most of the industry sells software that removes people from a process, Ethos sells software that puts them back in. Its tagline is almost a taunt: "You are so much more than what AI can automate." The company describes what it is building as "AI-powered infrastructure for human opportunity," and its two founders have spent the past two years turning that phrase into a working marketplace for expertise.

01 / What it isAn expert network, rebuilt from the phone up

To understand Ethos, start with the industry it is aiming at. Expert networks - firms like GLG, AlphaSights and Third Bridge - are a quietly enormous business. They connect investors and consultancies with people who know something specific: a former supply-chain lead at a chipmaker, a doctor who has prescribed a particular drug, an engineer who worked on a rival's product. A hedge fund pays for a one-hour call; the network takes a healthy cut; a broker somewhere spends the afternoon on the phone lining it all up.

Ethos keeps the premise - human knowledge, delivered on demand - and rips out the manual middle. Its onboarding is a voice agent: an AI that calls professionals and interviews them about what they have actually done, then reads their public trail (academic papers, GitHub repositories, podcasts, professional profiles) to fill in the rest. The output is a profile that models the "full shape" of a person's expertise rather than a job title. On the client side, a buyer describes a need in plain language, and Ethos matches it against a knowledge graph it says spans tens of millions of companies and millions of papers and podcasts.

02 / How it worksFrom a phone call to a paid engagement

The mechanics are best read as a pipeline. A professional joins, gets interviewed by voice, and lands in the graph. A client asks a question. The system finds the right people - and, increasingly, can conduct the interview itself and fold the answer into a deliverable.

That last step is the point. Ethos routes several kinds of paid work to its experts: consulting calls, market-research surveys, AI data-labeling and training projects, fractional roles and full-time jobs. It says its onboarding spans an unusually wide range of professions - accountants, bankers, consultants, lawyers, doctors, and skilled trades like plumbers and electricians. Top earners on the platform are reported to make more than $10,000 a month.

03 / The peopleA DeepMind researcher and a SoftBank operator

The founding pair is a deliberate two-body problem: research plus go-to-market. Daniel Mankowitz, the CTO, spent about six years as a research scientist at Google DeepMind, working on reinforcement learning and agentic systems - including AlphaDev, the project that discovered faster sorting algorithms now baked into standard code libraries, as well as work touching Gemini and video compression. James Lo, the CEO, came up through McKinsey and the SoftBank Vision Fund, where he worked on portfolio-company transformations, and previously founded and sold a conversational-AI company called Mana in 2023.

04 / The money$26M, and a fast clock

Ethos has moved quickly. It emerged from stealth in March 2025 with a seed round of roughly $3.25M led by General Catalyst. Fourteen months later, in May 2026, it announced a $22.75M Series A led by Andreessen Horowitz, with General Catalyst returning alongside XTX, Evantic Capital and Common Magic. That is about $26M in total. Within weeks of the round, the company shipped a new product and listed it on AWS Marketplace.

The a16z framing is worth quoting because it explains why a top-tier fund backed a headhunting-adjacent startup: "The last great improvement in the NPS of employment was Deel, making work global. The next will be Ethos: making work abundant." In other words, the investors are treating Ethos less as a research vendor and more as a future-of-work platform - a bet that AI can reprice human expertise more accurately than a resume ever could.

05 / The product betAn agent that makes the call for you

In mid-2026 Ethos launched the Ethos Investment Researcher, which it describes as an autonomous investment-research agent. The idea is aggressive: rather than returning a list of experts for a human analyst to chase, the agent plans a deep dive, sources and interviews relevant experts on the client's behalf, and returns an industry-standard report in minutes. It is the expert call, automated end to end - and it is available through AWS Marketplace.

This is where Ethos's contrarian thesis gets tested against itself. If an AI agent can do the research, why keep the human in the loop at all? The company's answer is that the hardest, most valuable part of research is not writing the memo - it is finding the right person and getting them to talk. Ethos automates the finding and the calling, and treats the human's tacit knowledge as the scarce asset worth paying for.

06 / The marketWhere it sits, and who it's up against

Ethos is threading between two markets that are both being reshaped by AI. On one side are the traditional expert networks - GLG, AlphaSights, Third Bridge, Guidepoint, Dialectica, and Tegus (now part of AlphaSense) - broker-led businesses whose economics rest on human sourcing and scheduling. On the other are the AI research and survey tools - Perplexity-style research agents, and platforms like Listen Labs and Outset - that automate synthesis but lack a vetted human network. And looming over both is LinkedIn, the default place people still go to find a person.

Ethos's wager is that the winning position is in the middle: an AI-native marketplace with a real supply of vetted humans and an agent that can actually reach them. Its business model reflects that - a marketplace take rate reported around 30% on the work it routes, with subscription access planned for its research products. The company is reported to be tracking toward eight-figure annualized revenue, though it has not published a valuation or a public price list, and some third-party data on its headcount and totals is inconsistent.

07 / What you can copyThe idea worth stealing

The transferable insight in Ethos is not the voice tech - it is the reframing. Most AI companies ask, "What work can we remove a human from?" Ethos asks, "What is a human uniquely worth, and how do we make that findable and priceable?" For anyone building in recruiting, research, or professional services, that is a useful inversion: the scarce input is not the report, it is the person who can speak to something the model has never seen.

The conditions where it might not hold are visible too. Automated expert networks live under real regulatory scrutiny - compliance and information barriers matter when hedge funds are on the line. A voice interview is only as good as the honesty of the person on the call. And the take-rate model depends on Ethos staying the shortest path between a question and the right human; if the incumbents bolt AI onto their existing supply, or if buyers get comfortable with a purely synthetic answer, the middle gets crowded. For now, Ethos is betting that the human on the other end of the line is the part that does not commoditize.

Figures for weekly expert sign-ups, earnings, headcount and revenue are as reported by the company and press coverage and are approximate. Funding totals combine a 2025 seed round and a 2026 Series A.