On October 20, 2026, at the Rhapsody Connect Executive Track in Boston, Rhapsody announced the general availability of Guardian, a healthcare-native gatekeeper for AI agent access to clinical data and services. It evolves the company’s earlier API Guardian, introduced in October 2025 – a distinct, earlier product – from static API management into runtime agent governance. The release’s shortest useful line: Guardian is available now.
This is a shipped control plane, not a roadmap.
The central tension is the gap between open-ended model capability and the rigid requirements of clinical environments. Generic identity providers like Okta’s agent identity framework authenticate users and increasingly agents, but they sit at the platform layer. Guardian sits at the domain layer: it authenticates agents, enforces scoped access and rate limits, and records agent, MCP, and API activity – the three mechanics named verbatim in the launch release. Rhapsody frames it as the first healthcare-native agent gatekeeper; that framing is positioning, worth noting and worth not repeating as category fact.
Mechanically, the notable move is what gets governed. The Model Context Protocol (MCP) is treated as a first-class governed endpoint alongside FHIR and REST interfaces – per the Guardian product page, an agent can be authorized to a designated service, restricted to permitted operations, and throttled on request frequency, with an audit trail of everything passing through. MCP is where agents call tools; treating it as a governed surface rather than trusted plumbing is the whole architectural bet.
For enterprise architects, the deeper signal is that the audit trail is becoming the product. In a regulated environment, the question is rarely can the agent do this?
but can we show, after the fact, what the agent was allowed to do and what it actually did?
Rhapsody’s own pitch says Guardian helps health systems put CHAI and Joint Commission guidance into practice
through access controls and audit trails – the vendor’s compliance framing, stated as such, and the detail that would prove it does not ride on this launch. What is verifiable today is narrower and still useful: scoped access, rate limits, and recorded activity, deployable standalone or alongside Rhapsody, Corepoint, or another integration platform.
Look past the launch and the pattern is structural. Governance is verticalizing. TGVP’s infrastructure thesis names governance as the gating layer for agent adoption at scale; platform-layer answers like Okta and context-layer answers like CAEVES’s managed context service each cover part of the surface. Healthcare’s part is different in kind: clinical data carries obligations that generic proxies do not model. With no shared agent-governance specification shipped – the specs under discussion at AGNTCon this week remain unratified – regulated verticals are not waiting. They are standing up their own gatekeepers, and the compliance machinery those verticals already run becomes the de facto standard before any formal one exists.
That is what this launch actually documents. Not a new category crowned, but a market past its first excitement: builders are now buying the operational realities of production – who controls the rails, what the agents may touch, and what the record shows afterward. Rhapsody CEO Sagnik Bhattacharya put the demand plainly: CIOs are under pressure to advance AI initiatives while maintaining control of sensitive healthcare data. They need to know which agents are accessing their systems, what those agents are allowed to do and whether they stay within those boundaries.
Guardian’s bet is that those boundaries, enforced and recorded inside a regulated vertical, are worth more than a generic gate.