Ethos raises $22.75M from a16z for voice-onboarded expert network

Ethos, a London-based startup building an expert network with voice-powered onboarding, raised $22.75M led by Andreessen Horowitz, according to TechCrunch. The platform uses conversational AI to interview experts and assemble structured profiles, then lets clients query that pool in natural language rather than by job title or keyword.

The pitch is squarely aimed at the workflow gap most expert networks acknowledge in private: resumes and self-tagged skills are a poor index of what someone actually knows. Ethos is betting that a 20-minute voice interview captures the texture (prior projects, niche operating experience, what the expert has actually shipped) that a LinkedIn profile flattens out.

The round size is modest by 2026 AI standards but notable for the category. Expert networks have historically been built on relationship-led sourcing (moderators who know which ex-Nike supply chain VP to call for a tariff question), and the venture market has largely funded the transcript and compliance layers around them rather than the matching layer itself. a16z writing a Series-style check into the matching problem is the signal worth tracking.

What Ethos appears to be selling, based on the source description, is two things bundled: a richer expert profile (voice-elicited, AI-structured) and a query interface that lets a research analyst describe the expert they need in a sentence rather than reverse-engineering Boolean search. The first is a data problem. The second is an interface problem. Both are real.

What the source covers, and what it doesn't

TechCrunch's piece, as summarized in the source, does not disclose Ethos's current expert count, customer list, revenue, or pricing model. It also doesn't address how the platform handles the compliance layer that defines the expert network industry: MNPI screening, exclusion lists for client compliance teams, and the audit trail that buy-side firms require before a call can be cited in an investment memo. Those omissions matter. A faster matching layer without those controls is a sourcing tool, not an expert network in the sense that GLG, Guidepoint, AlphaSense/Tegus, or Third Bridge sell it.

The source frames the buyer set as "consulting, finance, healthcare and enterprise strategy." Those four segments have very different tolerance for AI-mediated sourcing. Strategy consultants will adopt anything that compresses a two-week ramp into two days. Hedge fund analysts will not cite an expert in an IC memo if they can't trace the compliance chain back to a named human reviewer.

Why it matters

What to ask next: how many experts are on the platform today, what the average interview length is, whether profiles are refreshed, and what compliance controls (if any) sit between a client query and an expert connection. Those four answers determine whether this is a sourcing tool dressed up as an expert network, or a real attempt to rebuild the category.