Ethos raises $22.75M from a16z to scale its AI-driven expert network using voice onboarding

What Happened

Ethos, a next-generation expert network, has secured $22.75 million in funding led by Andreessen Horowitz (a16z). The company is leveraging AI-driven voice onboarding to rapidly scale its platform, reportedly adding 35,000 new experts to its network every week. This capital injection will accelerate their infrastructure development and market penetration against legacy incumbents.

Technical Details

The core technical differentiator for Ethos is its automated voice onboarding system. Traditionally, expert networks rely on human associates to identify, contact, and vet potential subject matter experts—a highly unscalable and labor-intensive process. Ethos replaces this with conversational AI agents capable of conducting natural language interviews. These agents utilize advanced speech-to-text (ASR), large language models (LLMs) for real-time contextual questioning, and entity extraction to map an expert's specific knowledge domains. By processing these voice interactions, Ethos can programmatically generate a highly structured, multidimensional knowledge graph of human expertise, converting unstructured conversational audio into normalized, queryable data points.

Why It Matters

From a data engineering perspective, Ethos is solving a massive data acquisition bottleneck. Human knowledge that isn't published online is notoriously difficult to index. By automating the extraction of this "tacit knowledge" via voice AI, Ethos is creating a proprietary dataset that cannot be easily replicated by standard web scraping. At an ingestion velocity of 35,000 experts per week, the compounding data moat is substantial. Furthermore, this validates conversational AI not just as a customer service tool, but as a primary ingestion pipeline for complex, high-value B2B data.

What to Watch Next

Monitor how Ethos handles automated vetting accuracy at this scale. The primary risk of algorithmic ingestion is the degradation of data quality, specifically the infiltration of fraudulent experts or exaggerated credentials. Additionally, watch for how they expose this knowledge graph to clients—whether through traditional manual matching, or via an LLM-powered search interface that allows clients to directly query the aggregated expertise of the network.