Cequence Shipped a Commercial Agent Identity Layer. The Law Has Not Caught Up.
The modern storefront is currently being overrun by invisible guests. Businesses are struggling to distinguish between a loyal customer and a bot acting on their behalf, a friction point that has turned into a full-blown operational crisis. As we move deeper into agentic commerce , the agent identity problem has shifted from a theoretical concern to a daily headache for CISOs. We have already seen how millions of users are shopping under a borrowed identity , and as the limit moves into the credential , the liability remains stubbornly stuck with the merchant.
Cequence is attempting to monetize this anxiety. On October 7, the company launched Agent Trust , a commercial capability integrated into its existing Application and API Protection platform. It is a pragmatic, if reactive, piece of plumbing. Rather than waiting for a centralized authority to define who or what an agent is, Cequence has built a system that tries to manage the chaos at the transaction layer.
The mechanics of the tool are designed to address the visibility gap. It uses a detected agents inventory powered by CQ Prime rules to flag autonomous traffic, while an agent activity log uses behavioral profiling to map how these bots interact with a site. Crucially, the system includes a protocol-agnostic issuer registry. This is the most interesting part of the build; it supports standards like Skyfire, Visa TAP, Google AP2, and OpenAI/Stripe ACP. By making the registry agnostic, Cequence avoids the trap of being tied to a single standard, allowing for new issuers to be registered without a full product overhaul. It is a clever way to hedge against the inevitable fragmentation of identity standards.
The legislative landscape offers no cover for these businesses. Senator Mark R. Warner introduced S. 5051, the AI AGENT Act, back in July 2026, proposing an FTC-administered registry and NIST-backed identity standards. The bill is currently languishing in the Senate Commerce Committee, leaving a vacuum where the responsibility for verification falls entirely on private infrastructure providers. Without a federal mandate, the burden of proof is being offloaded to the merchant.
This creates a binary trap: block the agent and lose the customer, or allow the agent and risk the security of the transaction. Cequence is pushing for a middle ground. Their platform allows for transaction-level enforcement, meaning administrators can rate-limit or challenge specific actions rather than blocking the entire agent. They have also added a biometric check feature, which forces the account holder to confirm on their own device that they actually intended for the agent to execute the transaction. It is a way of saying that if the machine cannot be trusted, the human must be brought back into the loop.
Cequence VP Hari Nair frames the utility of this approach with a blunt assessment: “AI agents like Meta’s Muse are bots that act on behalf of real customers. Many businesses can’t see them, and those that can are tempted to block them. Agent Trust lets businesses block the action, not the agent.” The data supports the urgency; at some Cequence customers, Muse traffic appeared at more than 50% of sites within two weeks, and at financial institutions, the bot was even completing MFA. For a travel customer, the noise was significant enough that they were blocking roughly one in five requests.
The broader strategy, according to Cequence CEO Ameya Talwalkar, is to normalize this integration. “Every CISO is about to hear the same question from the business, which is whether the company can let AI agents in. Saying no is no longer the safe answer, because customers are choosing agents and will take their spending wherever those agents are welcome,” Talwalkar noted. It is a compelling argument for adoption, though it glosses over the reality that this is a vendor-led solution to a systemic problem.
We are effectively watching the market build its own identity layer in real-time, a patchwork of private registries replacing what many hoped would be a government-led standard. The effectiveness of this approach will ultimately depend on whether these tools can reliably distinguish between a legitimate, user-authorized action and malicious automation. As the volume of agent-driven traffic scales, the industry is betting that this fragmented, vendor-led plumbing will hold. For now, it is the only option on the table.