Asymptote Labs Raises $5.5M to Put a Security Layer on AI Agents

The check from Tower Research Ventures was $5.5 million [The Next Web, October 2026]. The bet is that enterprise security teams will soon need a new category of software to see what their AI agents are doing. For Asymptote Labs, a New York startup founded this year, the seed round is a vote that the infrastructure for monitoring and controlling autonomous software is already overdue.

The wedge is visibility

AI agents are moving from demo to deployment, becoming an operating layer inside companies [PERPLEXITY SONAR PRO BRIEF]. The infrastructure to understand them, however, is still being written. Asymptote's initial product, an open-source tool called Beacon, is designed to be that system of record. It collects telemetry from agent activity across more than 20 different development harnesses, including Claude Code and Cursor, creating a shared history of their work [Welcome to Beacon - Beacon]. The company calls this an "AI Detection & Response layer for the agentic enterprise" [PERPLEXITY SONAR PRO BRIEF]. The wedge is simple: make agent actions visible and actionable before they become a security or compliance problem.

A team built on security precedent

The founders, Justin D’Souza and Shukan Shah, have worked together for over seven years [PERPLEXITY SONAR PRO BRIEF]. Their most recent shared stop was Doppel, a social-engineering-defense company that raised a $70 million Series C in late 2025 [Doppel Raises $70M Series C to Meet Rising Demand for AI-Driven Social Engineering Defense, November 19, 2025]. That experience in a security category that scaled quickly informs their approach. D’Souza, the CEO, led machine-learning engineering at Level and Caffeine. Shah, the CTO, worked on payments at Brex and Level before focusing on AI [PERPLEXITY SONAR PRO BRIEF]. Their pitch is not just about observability, but about building a security harness from the ground up.

Role	Name	Prior Experience
Co-Founder & CEO	Justin D’Souza	Machine-learning engineering at Level, Caffeine, Doppel
Co-Founder & CTO	Shukan Shah	Payments at Level, Brex; AI engineering at Doppel

The open-source to enterprise path

Asymptote is building as a progression, not a single monolithic product [Asymptote, Asymptote Overview]. The open-source Beacon is the foundation. From there, the company offers a managed enterprise platform and private deployment options for what it terms AI agent endpoint security [Asymptote Labs, Pricing]. The architecture uses an SDK to instrument agent applications, emitting OpenTelemetry data to a customer-managed collector or a hosted service [Asymptote, Open Source Architecture]. This model, familiar from companies like Datadog, aims to get developers using the tool first, with the security team becoming the eventual enterprise buyer.

The recent seed round, which was oversubscribed according to one report, included participation from South Park Commons, Lytical Ventures, and 1984 Ventures [Unite.AI, October 2026]. Tower Research Ventures is not just the lead investor; its own security organization is evaluating the technology on its endpoints [PERPLEXITY SONAR PRO BRIEF]. Asymptote says security teams are already using its tools in production, though it has not publicly named customers [The Next Web, October 2026]. The company is now hiring for five founding roles in New York, including a security engineer and a developer relations lead, with compensation listed between $150,000 and $300,000 [PERPLEXITY SONAR PRO BRIEF].

Where the wheels could come off

The bet is clear, but the market is still forming. The primary risk is timing: enterprise adoption of complex, multi-step AI agents may be slower than the infrastructure built for them. Asymptote must also navigate a landscape where large cloud providers or existing security platforms could decide to build similar observability features in-house. Furthermore, the company's success hinges on a dual adoption motion: developer love for the open-source tool, followed by security budget for the enterprise platform. Neither is guaranteed.

• Market timing. Widespread deployment of the heterogeneous agent fleets Asymptote is built for remains an emerging trend, not an established norm.
• Competitive encroachment. Existing application performance monitoring (APM) and security information and event management (SIEM) giants could extend their platforms into this space.
• The adoption funnel. Converting open-source users into paying enterprise customers is a classic, but difficult, go-to-market challenge.

For now, the $5.5 million seed led by Tower Research Ventures gives Asymptote Labs runway to build. The question for security chiefs is whether the first layer of defense for autonomous software will be a standalone platform, or a feature bundled into something they already own.

Sources

1. [The Next Web, October 2026] Asymptote Labs raises $5.5M to track and control what AI agents do | https://thenextweb.com/news/asymptote-labs-5-5m-seed-ai-agent-security
2. [PERPLEXITY SONAR PRO BRIEF] Company and product overview
3. [Welcome to Beacon - Beacon] Beacon product description
4. [Doppel Raises $70M Series C to Meet Rising Demand for AI-Driven Social Engineering Defense, November 19, 2025] Doppel funding announcement
5. [Unite.AI, October 2026] Asymptote Labs Raises $5.5M Seed to Monitor and Control Enterprise AI Agents | https://unite.ai/asymptote-labs-raises-5-5m-seed-to-monitor-and-control-enterprise-ai-agents/
6. [Asymptote, Asymptote Overview] Company building philosophy
7. [Asymptote Labs, Pricing] Product pricing and deployment options
8. [Asymptote, Open Source Architecture] Technical architecture details