Sycamore Raises $65M Seed to Build the Operating System for Enterprise AI Agents

Ex-Atlassian CTO. Backed by Coatue, Lightspeed, Bob McGrew, and Ali Ghodsi. The infrastructure layer every Fortune 500 needs to deploy agents safely.

April 3, 2026 · 7 min read · By Connie

The Problem Sycamore Is Solving

Enterprise AI agent adoption is accelerating faster than governance frameworks can keep up. Large companies are deploying dozens — sometimes hundreds — of AI agents across their operations: customer service agents, coding agents, data analysis agents, procurement agents. Each one takes actions, accesses systems, and makes decisions.

The question enterprises cannot currently answer is: what are all those agents doing, and can we trust them?

This is the gap Sycamore fills. The company describes its product as an "agentic operating system" — a platform layer that sits between a company's data and systems and the AI agents that interact with them. It handles four core functions:

• Discovery: Catalog every AI agent running in your organization, who built it, and what it can access
• Deployment: Controlled rollout of agents with access policies and integration with existing SSO and identity systems
• Observability: Real-time monitoring of what each agent is doing, what data it accessed, and what decisions it made
• Governance: Policy enforcement, human-in-the-loop escalation paths, and audit trails for compliance

The Team and the Round

Sri Viswanath brings unusual credentials to this problem. As CTO of Atlassian — the company behind Jira, Confluence, and Trello, used by millions of software teams worldwide — he spent years building enterprise infrastructure at scale with strict security and compliance requirements. He knows enterprise IT buying cycles, security team concerns, and what it takes to get approval from a Fortune 500 CISO.

The investor lineup reads like a who's who of enterprise AI infrastructure credibility:

How Sycamore Differs From Existing Agent Frameworks

The enterprise AI agent tooling landscape is crowded with frameworks — LangChain, Microsoft Semantic Kernel, AutoGen, CrewAI. These tools help developers build agents. Sycamore is positioned one layer above: it governs agents after they are built, regardless of which framework was used to create them.

Think of it as the difference between building a car (LangChain) and building traffic management infrastructure (Sycamore). One car doing whatever it wants is a feature. Ten thousand cars doing whatever they want is a city-wide problem.

This is exactly the problem Fortune 500 companies face: their developers build agents using any of a dozen frameworks, and the IT security and compliance teams have no visibility into what those agents are doing, what data they can access, or whether they can be audited.

What This Means for the AI Agent Market

Sycamore's raise is the latest signal that the AI agent infrastructure market is entering a consolidation phase. The question is no longer "can we build AI agents" — that's solved. The question is "can we safely scale them across a 50,000-person organization with auditable compliance."

For mid-market companies and teams that need AI agents today without $65M in infrastructure investment, platforms like Happycapy provide a more accessible entry point — hosted AI with built-in context management, conversation history, and multi-model support — while the enterprise-grade governance layer Sycamore is building matures.

Frequently Asked Questions