Asymptote Labs raises $5.5m for AI agent runtime security

Asymptote Labs has raised $5.5 million in seed funding to commercialise Beacon, a runtime security platform for AI agents, according to a report published by SecurityWeek on 21 October 2026. The round was led by Tower Research Ventures.

SecurityWeek said the money is intended to take Beacon, described as an AI agent runtime security platform, to market. Runtime security refers to controls that operate while software is actually executing, rather than checks applied to code or models before deployment. In the context of AI agents, that generally means monitoring what an agent does as it does it, as opposed to testing it in advance.

The report does not set out which agent frameworks or model providers Beacon supports, how it is deployed, or whether it is generally available. Nor does it name existing customers or give pricing. Asymptote Labs has not published further technical detail in the material reported.

Funding of this size at seed stage is modest by the standards of the wider security industry, and the round is an early bet rather than evidence of a proven product. Readers evaluating tools in this category should ask vendors directly for specifics: what telemetry the product collects from an agent, whether it can block an action or only record it, what latency it adds to agent execution, and how it behaves when an agent calls external tools or other agents.

The investment reflects continuing commercial interest in securing agentic systems, which are AI models given the ability to take actions such as calling APIs, browsing the web or running code. Because those actions happen quickly and often without a person reviewing each step, several vendors have argued that controls need to sit alongside the agent at execution time. Asymptote Labs is entering that market with Beacon.

The Agentic Times has not independently tested Beacon and has not verified the company's claims about its capabilities. Asymptote Labs did not publish a technical whitepaper alongside the funding announcement reported by SecurityWeek. Buyers should treat the launch as an early-stage offering and seek a proof of concept in their own environment before relying on it for production agent deployments.