AI Agents Are Becoming Economic Actors — and Marketplaces Have No Way to Verify Them

Baselayer, a San Francisco-based identity verification startup, raised $35 million in Series A funding to extend its business verification technology to AI agents. The company already serves over 2,000 financial institutions with Know Your Business (KYB) tools. Its new product — 'Know Your Agent' (KYA) — issues credentials to AI agents so merchants and platforms can verify whether an agent is authorized to act on behalf of a real person or business. It has partnered with FIS, Prove, and Socure to begin building adoption of this credential standard.

This is not a fintech story — it is a trust infrastructure story. AI agents are beginning to transact autonomously: booking services, making purchases, opening accounts, moving money. Marketplaces were built to match humans with humans. The entire trust layer — reviews, verification, reputation, fraud detection — assumes a human on both sides. That assumption is breaking. The deeper signal here is that a new category of marketplace participant is emerging — one with no history, no reputation, and no persistent identity. Traditional community marketplace best practices are built around human accountability, but the race to establish a verification standard for agents is, in effect, a race to own the trust layer of the next generation of commerce.

SUPPLY: AI agents are beginning to appear on the supply side of marketplaces — automating service delivery, responding to requests, executing tasks. Without verified identity, platforms cannot assess whether an agent is legitimate or a fraud vector. DEMAND: On the demand side, consumers and businesses are increasingly deploying agents to make purchases and book services on their behalf. Platforms have no mechanism to confirm the agent is authorized by a real buyer. LIQUIDITY: Unverified agents create friction that reduces liquidity. Platforms are already blocking bot-like behavior — but legitimate agents get caught in the same filters, suppressing real transactions. TRUST: The entire trust stack of a marketplace — verified profiles, reviews, ratings, dispute resolution — is built around persistent human identity. Agents have none of this. A single-task agent that 'spins up and spins down' cannot build reputation over time. GROWTH: Fraudsters are using AI agents to industrialize identity fraud — running applications 24/7 at scale. For marketplaces, this means fraud rates could spike sharply as the cost and effort of fraud collapse. ONBOARDING: KYB and KYC flows were designed for humans. AI agents acting as business representatives create a gap in onboarding logic — neither fully human nor fully machine, they fall outside current verification frameworks. MONETIZATION: Platforms developing a coherent AI marketplace operations strategy that cannot distinguish legitimate agent activity from fraud will either over-block (losing real revenue) or under-block (absorbing fraud losses). Either outcome damages take rate and unit economics.

• Audit your current onboarding and fraud detection flows for a simple gap: do they assume a human is always on the other side? If yes, document where that assumption breaks. • Review your terms of service to determine whether AI agents acting on behalf of users are permitted, restricted, or simply unaddressed. Close that gap now, before a dispute forces you to. • If you operate a service marketplace — where a provider fulfills a task — assess whether AI-generated supply is already entering your platform and whether your quality and trust systems can detect it. • Watch the credential standard race. Baselayer, and likely competitors, are building agent identity infrastructure. When a standard gains traction with financial institutions, evaluate integrating it before fraud pressure forces your hand. • For high-value or high-risk transaction categories on your platform, add an explicit 'human confirmation' checkpoint — even if temporary — until agent verification infrastructure matures. • Do not wait for the problem to be obvious. Fraud enabled by AI agents is cheap, fast, and continuous. The cost of being reactive is significantly higher than the cost of updating your trust framework now.