Dome Systems

Dome Systems is an enterprise software company building what it calls an agent operations platform: a single system of control for AI agents regardless of how they were built or where they run. Its stated premise is that every agent consists of three components — code (runtime and orchestration), a model (the reasoning engine), and tools (APIs, databases, messaging systems the agent calls) — and that each component is a governance surface. The product is organized around three corresponding modules: a Registry that makes each agent known and manageable by issuing credentials, declaring capabilities, and owning lifecycle; a Gateway that evaluates every tool call against policy with full context and can allow, deny, or escalate the call as well as filter returned data; and a Broker that constrains how an agent reasons through policy-based model selection, routing across providers, and evaluations against real traffic. Dome describes the tool call as the boundary every agent crosses and where most governance takes place.

The company frames the problem as agent proliferation and fragmentation: because the cost of producing working software has fallen sharply, every business function ships agents, each with its own framework, model, tool set, and identity scheme, and existing controls (RBAC, API gateways, audit logs, network policies) were designed for deterministic applications rather than probabilistic ones that decide at runtime what to do with their permissions. Dome addresses two deployment patterns with the same control plane: "AI In" (Agent Consumer — desktop assistants, SaaS-embedded agents, and internal builds used by the business) and "AI Out" (Agent Provider — agents the enterprise ships to customers and partners, including product surfaces exposed to agent callers).

Per its seed investor, Dome's workflow follows three operational principles — connect (assign identity, surface permitted tools filtered by policy, provision credentials, integrate via SDK, sidecar, or gateway regardless of runtime), secure (deterministic authorization evaluated at the tool-call boundary before an action executes), and operate (record every governed decision with who, what, under which policy, and with what result, streaming events into the enterprise's existing infrastructure rather than replacing it). The company also publishes long-form "perspectives" content, including an April 2026 paper on implementing platform services for AI agents in the enterprise.

Founding story

Dome Systems was founded by Dave McJannet (co-founder, CEO) and Marc Holmes (co-founder, COO). On the company's about page the founders state they started Dome because they believed the answer to enterprise agent adoption was a governance layer rather than tooling, arguing that the agent era needs the kind of foundational governance layer that previous computing eras eventually produced, purpose-built for the patterns of access and action specific to agents.

Business model

Dome sells an enterprise platform — a control plane layered above existing cloud, data, SaaS, and developer-tool surfaces — rather than replacing them; it integrates via SDK, sidecar, or gateway and streams governance events into the customer's existing infrastructure. Sources do not describe pricing or contract structure.

Traction

Dome is onboarding select enterprises into an early access cohort. Its careers page lists six open roles (four engineering Product Engineer positions and two enterprise Account Executive positions). No customer counts, revenue, or usage metrics are disclosed in the sources.

Latest developments

On 2026-04-16 Bessemer Venture Partners announced it had co-led Dome Systems' seed round, published by contributors Byron Deeter and Lauri Moore. On 2026-04-15 the company published a perspective paper, "Another Governance Layer, Again," outlining considerations for implementing platform services for AI agents in the enterprise. The company is actively recruiting engineers and enterprise sellers.