Baselayer raises $35M to verify the identity of AI agents

Baselayer raises $35M to verify the identity of AI agents

The company is launching "Know Your Agent" — cryptographic credentials for autonomous agents — at a time when no law requires anyone to check them. The M13-led round arrives in the middle of a race in which five competing products have launched in five weeks, none of them working together.

Identity-verification company Baselayer announced on September 22, 2026 a $35 million Series A, led by M13 with participation from Picus Capital, Torch Capital, Afore Capital and Matt Thompson of verification company Socure. Total funding now stands at $40 million, according to co-founder and CEO Jonathan Awad. The company has not disclosed a valuation. Alongside the round, Baselayer is launching its Agentic Identity Suite — a product family meant to give AI agents what neither KYC nor KYB provides them: something concrete to hand over at the door of a bank or an online store.

The gap in today's rules

The premise is simply stated: as TechTimes points out, no one is currently required to check what an agent is authorized to do (TechTimes, 9c8c8691). The verification regimes that banks and payment companies build on assume a persistent counterparty. KYC, anchored in the Bank Secrecy Act, is about knowing people. KYB is about knowing businesses. An AI agent created for a single task and then shut down fits neither category.

Awad described the problem this way to Crunchbase: "Agents spin up, and they spin down. How can you trust this random one-task agent?" (Crunchbase News, f58ef154). An agent with no history gives the bank nothing to assess — no credit history, no registration number, no signature from a responsible person. Baselayer's answer is the concept it calls "Know Your Agent," or KYA: a system meant to establish not just who deployed the agent, but whom the agent represents, and whether it actually has permission to perform this specific task.

How it works technically — as far as the sources go

The Agentic Identity Suite builds on the company's existing KYB infrastructure and uses W3C Verifiable Credentials: cryptographically signed, time-limited and scope-limited tokens that bind an agent to the platform it is deployed from, and to the business authorizing it (Forkast News via Yahoo Finance, 2908fdc5). A trading partner receiving such a credential can verify the signature, check that it has not expired, and read what action authority it covers — without having to trust claims made by the agent itself.

Co-founder and CTO Timothy Hyde described the motivation in the company's own announcement materials (Finextra republishes them unedited as external content): "Agents don't carry ID, and the infrastructure built to verify humans and businesses simply doesn't recognize them. Static fraud checks end up blocking good customers while sophisticated attacks go straight through. We built the Agentic Identity Suite so that any institution can know, cryptographically, which agent it's dealing with, whom that agent represents, and whether its own agent can safely proceed with a transaction" (Finextra, d9c972bc).

It is worth noting that the sources do not give a complete picture of the technical architecture around credential issuance and validation; the description above is what the documentation actually supports.

The numbers — and why they require caveats

Baselayer says that over 2,000 US financial institutions — over 20 percent of such institutions in the US — use its technology for onboarding, credit assessment and merchant account opening. The company also says its customers have prevented more than $1 billion in fraud, that it works with Fortune 500 companies, and that it has around 50 employees in San Francisco and New York (Crunchbase News, f58ef154).

All of these figures are the company's own claims, not independently verified. TechTimes notes that neither the fraud figure nor the revenue figure has been audited, and Awad declined to provide revenue numbers. A further reader caution is warranted: several of the coverages — TechTimes, Pulse 2.0 and Finextra — republish the same press release, and Finextra labels the text as unedited external content. They therefore do not constitute independent confirmation of one another.

The same applies to the context Baselayer itself highlights: in its announcement, the company cites Stripe's report that 70 percent of commands accessing data via Stripe's API came from AI agents (Pulse 2.0, 429e61be). That figure is company-reported and dated inconsistently across the sources — from June 2026 to a narrower CLI-based figure from 2025 — so its scope and date should be treated as unresolved. Baselayer also points to agent-commerce initiatives from Visa, Mastercard, American Express and Shopify.

A market growing ahead of the rules

The timing is no accident. Forkast News' analysis, syndicated via Yahoo Finance, notes that no fewer than five competing agent-identity products have launched in five weeks — without interoperating with one another (Forkast/Yahoo Finance, 2908fdc5). The same analysis estimates that only around 3 percent of transactions today involve agents. That means an entire infrastructure industry is being built in anticipation of the flood, not in the aftermath of established use.

Baselayer is trying to position itself in the standards work rather than only in the product race. According to the company, via Ventureburn's coverage, it participates in the FIDO Alliance's Authentication Working Group and the x402 Identity Working Group — alongside, among others, Cloudflare, Google, Visa and Mastercard. Partnerships include FIS, Prove, Socure, Exa and Parallel Web Systems (Ventureburn, d9863753). These are also company-reported facts, but they point to what may really prove decisive: a credential is only worth something if the recipient can read it.

The unresolved questions

The hardest questions are not technical. Forkast frames merchant resistance as a liability question: according to the analysis's own earlier figures, only 23 percent of consumers trust AI to handle payment transactions, 93 percent of merchants believe the AI vendor should bear the financial loss from erroneous purchases, and only 28 percent are willing to offer agents their full product catalog. Forkast provides no methodology for these figures, so they should be read as directional indications, not documented measurements (2908fdc5).

Who is actually financially liable when an authorized agent transacts wrongly therefore remains open — and no regulator has so far required anyone to verify an AI agent at all. That means the credibility of KYA offerings like Baselayer's comes not from the law or from a funding round, but from whether the industry converges on interoperable standards. That is where Baselayer is already working — side by side with the large payment networks it otherwise has to convince.

Source basis: The article rests on secondary sources that largely republish the company's own statements; no primary source from Baselayer is available. The founding date (February 2023 according to TechTimes and Crunchbase, 2024 according to Pulse 2.0 and Finextra) and headquarters (New York and/or San Francisco) are reported inconsistently across the sources and are handled here as "offices in both cities."