Asymptote Labs: Runtime Security and Observability for AI Agents
Asymptote Labs is an AI infrastructure and security startup building tools to monitor, govern, and improve AI agent activity across enterprise environments. Its open-source project, Beacon, collects and standardises agent telemetry across different tools and runtimes. The company’s enterprise platform adds real-time security detections, policy enforcement, and controls to help organisations understand what their AI agents are doing and prevent risky actions. Asymptote announced a $5.5 million seed round on October 20, 2026.
Visit Official Website ↗Asymptote Labs at a Glance: Company Overview
Asymptote Labs builds runtime infrastructure for AI agents. Its platform is designed to give enterprise security, engineering, platform, and finance teams a shared view of how AI agents operate across development tools, workstations, cloud environments, and business workflows.
The company addresses a growing operational challenge: organisations may deploy AI agents across multiple tools and environments, but their activity can be scattered across separate logs and systems. This makes it difficult to identify risky behaviour, investigate failures, understand resource consumption, and enforce consistent policies.
Asymptote’s open-source Beacon project collects agent activity and normalises it into a common telemetry model. Its enterprise platform builds on that data layer with managed monitoring, security detections, policy controls, and governance workflows. The company initially focuses on runtime security while positioning its technology as broader infrastructure for managing AI agents at scale.
| Company Name | Asymptote Labs |
| Industry | AI infrastructure, cybersecurity, AI agent observability and governance |
| Founded | Publicly confirmed founding year not specified in the funding announcement |
| Founders | Justin D’Souza and Shukan Shah |
| Leadership | Justin D’Souza, Co-Founder and CEO; Shukan Shah, Co-Founder and CTO |
| Headquarters | New York City, United States |
| Core Product | Runtime infrastructure for AI agent visibility, security, and control |
| Open-Source Product | Beacon, also referred to as Agent Beacon |
| Enterprise Platform | Asymptote Platform |
| Primary Use Cases | Agent telemetry, runtime security, policy enforcement, activity monitoring, and performance optimisation |
| Latest Funding Round | $5.5 million seed round, announced October 20, 2026 |
| Lead Investor | Tower Research Ventures |
| Other Investors | Lytical Ventures, 1984 Ventures, and South Park Commons |
| Valuation | Not publicly disclosed |
| Security Compliance | The company reports SOC 2 Type II compliance |
| Official Website | asymptotelabs.ai |
| Open-Source Repository | Agent Beacon on GitHub |
What Does Asymptote Labs Do?
Asymptote Labs helps organisations understand and control AI agents while they are running. Rather than focusing only on the underlying AI model or the application that uses it, the company focuses on runtime activity: prompts, tool calls, file changes, approvals, and other events generated during agent sessions.
Monitors AI Agent Activity
Beacon collects activity from supported AI agents and developer tools. It standardises these events so teams can investigate behaviour across different environments without relying entirely on separate logs for each tool.
Detects Risky Agent Behaviour
Asymptote’s enterprise platform builds on runtime telemetry to identify potential threats such as data leakage, insider threats, and AI supply chain risks. Its controls are designed to help security teams respond before unsafe actions cause harm.
Improves Agent Reliability and Efficiency
The platform also helps engineering and platform teams investigate failures, understand agent usage, and identify inefficient workflows. Finance teams can use runtime data to understand resource consumption and spending.
Asymptote Labs Products and Key Capabilities
Asymptote uses an open-source telemetry layer as the foundation for its enterprise product. Beacon captures and standardises agent activity, while the managed platform adds centralised search, detection, governance, and runtime controls.
1. Beacon Open-Source Telemetry
Beacon is an MIT-licensed open-source project that captures activity from supported AI agents and normalises it into a common runtime data model extending the OpenTelemetry standard. It helps organisations maintain a consistent record of agent behaviour across tools and environments.
2. Runtime Visibility
The platform provides a unified view of agent sessions, prompts, tool calls, file edits, approvals, and runtime events. This visibility helps security and engineering teams investigate what an agent did and identify where a process failed.
3. Security Detection and Response
Asymptote’s enterprise platform includes detections for potential data leakage, insider threats, AI supply chain risks, and inefficient agent workflows. The platform is designed to surface findings and support response while agents are in use.
4. Runtime Policy Enforcement
The enterprise platform can apply policy-driven guardrails at the prompt and tool levels. These controls are intended to steer agents away from unsafe behaviour and block risky actions before they are carried out.
5. Managed Agent Telemetry
The managed platform adds enterprise capabilities for ingesting, retaining, searching, and analysing agent activity. Organisations can use these capabilities without building every monitoring and data management component themselves.
6. Agent Performance and Cost Analysis
By consolidating agent activity, Asymptote aims to help teams understand where agents are failing, which workflows consume resources, and where performance or spending can be improved.
How Asymptote Beacon Works
Beacon provides a common data layer for observing AI agents. Different agents and development environments generate different kinds of events, making direct comparison and investigation difficult. Beacon collects supported events and converts them into a more consistent representation that can be analysed by security and engineering systems.
Step 1: Capture Agent Activity
Beacon collects supported runtime events from tools such as Claude Code, Codex, Cursor, and other agent harnesses. Depending on the integration, these events can include prompts, tool calls, commands, file changes, approvals, and session activity.
Step 2: Standardise the Data
Beacon normalises events into a common runtime data model that extends the OpenTelemetry standard. This makes agent activity easier to correlate across supported tools and environments.
Step 3: Build a Shared View
Teams can use the resulting telemetry to investigate agent sessions and understand activity across users, tools, and environments. The open-source layer can also route telemetry into compatible systems.
Step 4: Identify Potential Risks
The enterprise platform analyses runtime data to surface potentially unsafe behaviour, data leakage, insider threats, and other security findings.
Step 5: Enforce Policies
Enterprise controls can apply guardrails at prompt and tool levels, helping block risky actions or steer agents towards permitted behaviour.
Step 6: Improve Operations
Security, engineering, platform, and finance teams can use agent telemetry to investigate failures, review usage, understand costs, and improve workflows over time.
Asymptote Labs Use Cases Across Enterprise Teams
As organisations adopt more AI agents, multiple teams need visibility into how those agents behave. Asymptote’s runtime data layer is designed to support security monitoring, software development, platform governance, and cost analysis from a shared record of agent activity.
Cybersecurity Operations
- Investigate potentially unsafe agent actions.
- Detect possible data leakage and insider threats.
- Identify AI supply chain risks.
- Apply runtime policies to supported agent activity.
Software Engineering
- Trace agent tool calls and file changes.
- Investigate failed coding-agent sessions.
- Understand the sequence of actions in a task.
- Identify inefficient agent workflows.
Platform Engineering
- Maintain visibility across agent deployments.
- Standardise telemetry from supported runtimes.
- Manage agent adoption across teams.
- Integrate agent data with existing infrastructure.
IT and Enterprise Operations
- Understand how agents interact with company systems.
- Investigate operational failures.
- Review agent activity across environments.
- Apply consistent governance controls.
Finance and FinOps
- Understand resource consumption by agents.
- Investigate inefficient workflows and repeated tasks.
- Analyse usage patterns across teams.
- Support cost reviews using runtime activity data.
Regulated Enterprises
- Improve visibility into agent interactions with sensitive systems.
- Maintain records for security investigations.
- Apply policy controls to supported agent actions.
- Review deployment and data handling requirements.
What Makes Asymptote Labs Different?
Asymptote focuses on runtime data and control for AI agents rather than only on model evaluation, conventional application monitoring, or endpoint security. Its approach combines open-source telemetry with an enterprise platform that can use that data for security detections and policy enforcement.
| Area | Traditional Approach | Asymptote Labs Approach |
|---|---|---|
| Agent Visibility | Agent activity may be spread across separate logs and tools. | Beacon normalises supported agent activity into a common runtime data model. |
| Integration | Teams may need to interpret different event formats for each agent environment. | Beacon aims to provide a consistent telemetry layer across supported agent harnesses. |
| Security Monitoring | Existing security tools may not provide complete context about AI agent actions. | Asymptote builds detections around agent runtime data and potential AI-specific risks. |
| Policy Enforcement | Controls may be distributed across individual tools and integrations. | The enterprise platform is designed to enforce guardrails at prompt and tool levels. |
| Data Ownership | Telemetry may be tied to a particular monitoring vendor or implementation. | The open-source Beacon layer is designed to give organisations a portable record of agent activity. |
| Operational Analysis | Security, engineering, and finance teams may investigate different datasets. | A shared runtime data layer can support investigation, governance, reliability, and cost analysis. |
These distinctions describe Asymptote’s product positioning. Beacon and the enterprise platform do not automatically replace every existing observability or security product. Organisations should evaluate integration coverage, deployment requirements, event retention, policy enforcement, and the overlap with their current tools.
Beacon and the Asymptote Enterprise Platform
Asymptote separates its open-source telemetry foundation from the managed capabilities designed for enterprise use. This gives development teams a way to inspect and collect agent activity while providing organisations with an option for centralised monitoring and control.
Beacon: Open-Source Visibility
Beacon is designed for teams that want to capture and inspect agent activity, route telemetry to compatible security systems, and maintain a common record of runtime events. Its MIT licence allows teams to examine and adapt the open-source project within the licence’s terms.
Asymptote: Managed Control
The enterprise platform adds managed ingestion, retention, search, detections, identity mapping, policy workflows, and centralised governance capabilities. It is aimed at organisations managing agents across teams and environments.
Private Deployments
Asymptote also offers enterprise and private deployment options. Organisations evaluating these options should confirm current hosting, data residency, integration, and support requirements directly with the company.
Asymptote Labs Funding History and Investors
Asymptote Labs announced an oversubscribed $5.5 million seed round on October 20, 2026. Tower Research Ventures led the round, with participation from Lytical Ventures, 1984 Ventures, and South Park Commons. The company said it would use the capital to expand engineering, product development, hiring, and go-to-market activities.
The company says Beacon launched in May 2026 as an open-source telemetry project for AI agents.
An oversubscribed seed round led by Tower Research Ventures, with Lytical Ventures, 1984 Ventures, and South Park Commons participating.
The company plans to invest in engineering, product development, hiring, and go-to-market. A valuation was not disclosed.
Angel backers named by the company include Doppel co-founders Kevin Tian and Rahul Madduluri, along with Socket founder Feross Aboukhadijeh. The funding announcement did not disclose a company valuation or a more detailed breakdown of the capital allocation.
Asymptote Labs Founders and Team
Asymptote Labs was co-founded by Justin D’Souza and Shukan Shah, who previously worked together at Doppel, a company focused on social engineering defence. Their experience in security detection, machine learning infrastructure, and security operations automation informs Asymptote’s focus on observing and controlling AI agents at runtime.
Justin D’Souza, Co-Founder and CEO
D’Souza previously worked on machine learning infrastructure at Doppel. According to Asymptote, he helped scale its detection platform from processing one million to more than one billion domains per day. His background informs the company’s emphasis on real-time data processing and detection.
Shukan Shah, Co-Founder and CTO
Shah previously worked on security operations automation at Doppel and developed agent-based approaches to threat hunting and investigation. His experience informs the company’s work on AI agent runtime infrastructure and security controls.
Security Research Experience
The founding team also draws on experience in security research and infrastructure engineering. Asymptote’s background is relevant to its focus on detecting risky behaviour, tracing agent actions, and developing controls for enterprise deployments.
Security, Privacy, and Enterprise Considerations
AI agent monitoring can involve sensitive information, including prompts, commands, file changes, and tool responses. Before deploying a runtime telemetry platform, organisations should understand what information is collected, where it is stored, who can access it, and how long it is retained.
Telemetry and Data Ownership
Beacon is designed to give organisations a portable record of agent activity. Enterprises should still review event contents, redaction options, storage locations, and data routing before enabling collection across employee devices and development environments.
Runtime Controls
Asymptote’s enterprise platform supports policy-driven guardrails at prompt and tool levels. Teams should verify which controls apply to their specific integrations and how blocked actions, exceptions, and approvals are handled.
Security Compliance
Asymptote reports SOC 2 Type II compliance. Organisations should request the current attestation and review its scope, data processing terms, access controls, and retention practices as part of vendor assessment.
Frequently Asked Questions About Asymptote Labs
What is Asymptote Labs?
Asymptote Labs is an AI infrastructure and security company that builds runtime visibility, detection, and control capabilities for AI agents. Its open-source Beacon project collects and standardises agent activity, while its enterprise platform adds monitoring, policy enforcement, and governance.
How much funding has Asymptote Labs raised?
Asymptote Labs announced a $5.5 million seed round on October 20, 2026, led by Tower Research Ventures. The announcement did not disclose a company valuation.
Who invested in Asymptote Labs?
The company’s seed round was led by Tower Research Ventures, with participation from Lytical Ventures, 1984 Ventures, and South Park Commons. Angel backers include Doppel co-founders Kevin Tian and Rahul Madduluri and Socket founder Feross Aboukhadijeh.
Who founded Asymptote Labs?
Asymptote Labs was co-founded by Justin D’Souza, CEO, and Shukan Shah, CTO. Both previously worked at Doppel, where they gained experience in security detection, machine learning infrastructure, and security operations automation.
What is Beacon by Asymptote?
Beacon, also called Agent Beacon, is an MIT-licensed open-source telemetry layer that captures and standardises activity from supported AI agents. It helps organisations build a consistent record of prompts, tool calls, file changes, approvals, and other runtime events.
Which AI agents does Beacon support?
Asymptote identifies Claude Code, Codex, Cursor, and more than 20 additional agent harnesses among the environments supported by Beacon. The exact coverage depends on the agent and integration, so teams should check the current project documentation before deployment.
How does Asymptote help secure AI agents?
Asymptote’s enterprise platform uses runtime telemetry to identify potential risks such as data leakage, insider threats, and AI supply chain issues. It also supports policy-driven guardrails at prompt and tool levels to help prevent unsafe actions.
Is Beacon open source?
Yes. Asymptote identifies Beacon as an MIT-licensed open-source project. Organisations can review its code and use it within the terms of the licence. Enterprise monitoring and control capabilities are offered through Asymptote’s separate platform.
How is Asymptote different from traditional observability tools?
Asymptote focuses on AI agent runtime events and their security implications. Beacon normalises activity across supported agent environments, while the enterprise platform adds AI-focused detections and policy controls. It can complement existing observability and security tools rather than automatically replacing them.
Who can use Asymptote Labs?
Its platform is designed for enterprise security teams, software engineers, platform teams, and finance teams that need to understand agent behaviour, investigate failures, enforce policies, or analyse usage and cost.
Is Asymptote Labs SOC 2 compliant?
Asymptote reports SOC 2 Type II compliance. Organisations considering the platform should request its current attestation and verify the scope and relevant controls during vendor assessment.
What is Asymptote Labs’ valuation?
Asymptote Labs did not publicly disclose its valuation in its October 20, 2026 seed funding announcement.
What is the official website of Asymptote Labs?
The official website is asymptotelabs.ai. Its open-source telemetry project is available through the Agent Beacon GitHub repository.