The Expert Network Is Being Eaten by the Same AI Wave It Was Supposed to Service
Ethos says its AI verification works. This reporter tested it.
On a Thursday afternoon, I named my expertise — machine learning research, foundation model capability evaluation, the kind of work a beat reporter spends years building intuition around — and requested an Ethos profile. The onboarding starts with a domain check: prove you have the expertise you're claiming. Ethos asks for a voice interview, a structured conversation with an AI that builds a richer profile than a CV can offer: academic papers, code repositories, conference talks, professional content, all ingested alongside the conversation. It then matches those profiles against client needs for consulting engagements, expert calls, market research, AI data-labelling projects, and full-time roles. The company takes thirty percent or more per project. Roughly 35,000 people join per week, though Ethos sends the invitations — meaning it controls who is in the denominator. TechCrunch
Forty-eight hours later, the invite arrived.
What that gap reveals is the business thesis in miniature. Generative AI has, in roughly thirty months, made it dramatically easier to look qualified for a job and dramatically harder for an employer to tell whether you actually are. The candidate-side tools scaled fast. The recruiter-side tools did not. Ethos is a position in a market that AI broke and is now trying to fix — and the buyers most motivated to pay for that fix are the same companies fastest at making it redundant. TechCrunch
The co-founders are James Lo (CEO, ex-McKinsey and SoftBank Vision Fund) and Daniel Mankowitz (CTO, ex-Google DeepMind research scientist who worked on YouTube's video compression algorithm, Gemini, and the AlphaDev sorting algorithm). That is a real combination: a DeepMind systems-design brain paired with McKinsey-and-Vision-Fund commercial discipline. a16z has historically liked that pairing for enterprise AI bets. The $22.75 million Series A was led by Andreessen Horowitz with participation from General Catalyst, XTX Markets, Evantic Capital, and Common Magic. Total funding is around $30 million including a $3.25 million seed round last year from General Catalyst. TechCrunch Tech.eu
Top experts on the platform earn over £7,000 per month; the average earner makes £4,500 per month, according to Ethos. The company is on track for eight-figure annualized revenue — a target, not a disclosed figure. TechFundingNews
The structural tension is in what Ethos is selling and who is buying. GLG, Guidepoint, and AlphaSights built billion-dollar expert network businesses over decades by curating rosters of retired executives, domain specialists, and consultants available for paid calls. Investment banks, law firms, and consulting companies paid premium rates for verified expertise. The model worked because that expertise was scarce and expensive. The Next Web
AI disrupted that in two directions at once. First, it made it easier for anyone to appear to have expertise they do not possess. Second — and less noticed — frontier AI labs are now spending to systematically map every economically valuable occupation, converting human expertise into training data at scale. The CEO of Ethos said as much on the record: AI labs are pointing a capital gun at every economically valuable occupation and trying to map out every profession. He called it an amazing tailwind for his business. TechCrunch
Here is the part that deserves scrutiny: GLG and Guidepoint rosters have themselves now been signed up as data partners inside Claude Opus 4.7. The expert networks that Ethos competes with are simultaneously being absorbed into the training pipelines of the AI labs most likely to disintermediate them. Ethos is selling talent verification to AI labs. The AI labs are simultaneously building systems that make human verification less necessary. If a frontier model can interpolate expert knowledge from training data scraped from the same expert networks Ethos is building, the marginal value of any single expert on the platform decreases. Ethos may be building the infrastructure for its own disintermediation — selling the human layer to the companies fastest at rendering that layer redundant. The Next Web
a16z has a pattern of making large early-stage bets on enterprise AI infrastructure that sound compelling in pitch decks and prove harder in the field. That is not a criticism of Ethos specifically. It is the epistemic frame the reader deserves when a company with eight employees raises $22.75 million on a story about AI verifying human expertise at scale.
The real question is not whether the DeepMind co-founder pedigree is real. It is. It is not whether the funding is real. It is. The question is whether verified human expertise is a durable asset in a world where the buyers are actively working to make it redundant, and whether Ethos can build enough switching cost into its platform before the labs finish mapping the expertise directly.
That is a bet. The press release calls it a Series A.