By Pietro Decio, Frank Odom, Jared Young
AI agents are taking on increasingly important tasks in modern enterprises. They write software, execute commands autonomously, and interact broadly with internal business systems. Most enterprises today still struggle to fully understand and oversee their AI activity. At best, teams have to piece together incomplete records of agent activity across different tools. At worst, significant portions of that activity go entirely unseen.
Much of this challenge stems from the growing fragmentation of models, agent harnesses, and compute environments. Engineers might switch between Claude Code, Cursor, and Codex depending on the task or latest model release. Knowledge workers might use browser-based interfaces like ChatGPT and Claude, agent harnesses like OpenClaw, or personal agents like Muse. Meanwhile, agents run across local machines, cloud environments, CI pipelines, and internal applications. Each tool maintains its own history in its own format.
Enterprises need a system of record for agentic work that covers every agent and captures the full context of each session. This is what Asymptote Labs is building. A single shared view of every agent, wherever it runs, so teams can observe behavior, impose controls on it, and optimize for effectiveness.
We are excited to announce that Tower Research Ventures led Asymptote’s seed round, alongside South Park Commons, Lytical Ventures, 1984 Ventures, and a group of security and engineering leaders from the likes of Cursor, Vercel, Doppel, Microsoft, and more.
Co-founders Justin D’Souza and Shukan Shah have spent more than seven years building together across multiple startups. At Doppel, Justin scaled machine-learning systems powering threat detection to over a billion domains per day. Shukan built agent-driven security automation for threat hunting and investigations. They’re now bringing that experience to Asymptote, assembling a world-class technical team that includes security research talent from Trail of Bits to tackle one of the defining infrastructure challenges of the agentic era.
Asymptote’s product foundation is Agent Beacon, an open-source runtime data layer for AI agents. Beacon captures activity across more than 20 agent harnesses, along with browser chat, cloud agents, and agent SDKs. It unifies prompts, tool calls, commands, file edits, approvals, and token usage in a common data model.
This gives enterprises a unified, cross-platform record of what their agents are doing and the context behind their actions. Companies can stream this telemetry into existing security and observability platforms like Splunk, Datadog, and CrowdStrike, while using the same data to investigate risks, understand agent performance, track costs, and improve how agents work. Rather than leaving valuable agent history fragmented across vendors, Beacon gives enterprises a shared data foundation they own and can build upon.
We at Tower first found the Asymptote team through Beacon. Our first-hand experience with AI telemetry gave us the conviction to invest.
Asymptote’s commercial platform leverages Beacon’s telemetry to offer enterprises a control plane for agent activity. It enables managed detections, as well as comprehensive search and retention. It also ties every action to the actor behind it, and lets teams put additional security checks in front of sensitive actions, such as an agent deleting production data. Beacon captures the record of agent activity; Asymptote’s platform enables enterprises to understand it and act on it.
This same record serves teams beyond security; it can show engineering why agents are failing, and finance teams where their token spend is going. It can be used to make the agents themselves more performant by sharing context and memory. Every session holds knowledge about a company’s systems and processes. Today most of it disappears when the session ends or stays locked inside one harness. Beacon extracts it and makes it available to future agents across tools, so a fix found in Cursor can help an agent in Codex. Asymptote’s core belief is that agent session history should belong to the company, not the harness. Giving companies ownership over this record lets its value compound with every session and every agent, regardless of what tool is used next.
We first discovered Asymptote through Beacon. What stood out was not just the technology, but the team’s understanding of how agents operate, where they fail, and what enterprises will need as adoption accelerates. We believe the record of agentic work will become one of the most valuable assets an enterprise owns, and that Asymptote is uniquely positioned to build the platform that makes it useful. We’re thrilled to partner with Justin, Shukan, and the team as they build it.
Learn more at asymptotelabs.ai. The team is also actively hiring! If you’re interested in building the future of agentic infrastructure, reach out at founders@asymptotelabs.ai.
The views expressed herein are solely the views of the author(s), are as of the date they were originally posted, and are not necessarily the views of Tower Research Ventures LLC, or any of its affiliates. They are not intended to provide, and should not be relied upon for, investment advice, nor is any information herein any offer to buy or sell any security or intended as the basis for the purchase or sale of any investment. The information herein has not been and will not be updated or otherwise revised to reflect information that subsequently becomes available, or circumstances existing or changes occurring after the date of preparation. Certain information contained herein is based on published and unpublished sources. The information has not been independently verified by TRV or its representatives, and the accuracy or completeness of such information is not guaranteed. Your linking to or use of any third-party websites is at your own risk. Tower Research Ventures disclaims any responsibility for the products or services offered or the information contained on any third-party websites.