Agentic artificial intelligence promises to revolutionize enterprise workflows, but its mass adoption brings a critical challenge: how to govern autonomous agents that interact with multiple systems? Snowflake has taken a strategic step by announcing the acquisition of Natoma, a startup specializing in the Model Context Protocol (MCP) standard, to offer enterprises a layer of control, security, and auditing over their AI agents. This move is not just technical but responds to an urgent need for CIOs: preventing agentic AI from becoming a new risk focus.

What Does Natoma Bring to Snowflake?

Natoma has positioned itself as a provider of MCP-based tools that securely connect AI agents with enterprise systems. Its platform offers governance, observability, and access control, elements that Snowflake will integrate into its ecosystem to enhance Cortex Agents, Snowflake Intelligence, and Cortex Code. In essence, Natoma acts as a control fabric ensuring agents only access what they should, under defined policies and with full auditability.

As Phil Fersht, CEO of HFS Research, points out, “MCP is becoming the connective tissue of enterprise agents, but without identity, policies, privileged access controls, and audit capability, it can quickly become a shadow AI risk.” This warning resonates strongly in a context where security in AI agents is increasingly critical, as demonstrated by the case of Gavriel Cohen and OpenClaw.

The MCP Challenge: Not a Plug-and-Play Miracle

Although MCP standardizes connections between agents and systems, it does not solve governance problems by itself. Robert Kramer, managing director at KramerERP, warns: “MCP is a protocol, not a governance model in itself. It can standardize connections, but it can also standardize risk if access is too broad, tools are poorly governed, or agents are trusted too quickly.”

Therefore, the acquisition of Natoma precisely aims to fill that gap: providing a governed MCP with verified servers, identity-based authorizations, policy enforcement, and audit capability. However, analysts agree that most enterprises are not yet ready for mass MCP adoption. “They want the productivity and context benefits, but their governance, identity, data classification, and access control models are still catching up,” says Fersht.

The Race for the AI Control Plane

Snowflake is not the only one seeking to position itself in this space. SaaS providers like Salesforce, ServiceNow, and Workday are integrating agentic orchestration capabilities, while hyperscalers like Microsoft, AWS, and Google consolidate their development tools. As Michael Ni, analyst at Constellation Research, notes, “Data platforms won the analytics era. Whoever governs agents, context, and autonomous actions will win the agentic era. Natoma brings Snowflake the missing layer between insight and execution.”

This trend is also reflected in initiatives like Project Lightwell, where IBM and Red Hat bet on enterprise open source, or in the development of autonomous agents like Skipper, which deploys without asking for opinions. The key difference now is governance: without it, automation can become a risk.

Implications for CIOs

For CIOs, Snowflake's acquisition of Natoma sends a clear signal: agent governance must be a priority. Critical points to watch include identity-based permissions, least privilege access, audit logs, human oversight for high-risk actions, data leakage controls, and clear ownership when an agent makes a bad decision.

In this regard, process automation with tools like n8n and AI can benefit from these control layers, but also requires companies to establish solid policies from the start. As Fersht warns, “CIOs must avoid treating MCP as a plug-and-play miracle.”

The Future of Governed Agentic AI

The integration of Natoma into Snowflake promises to simplify agent management at scale, but the real challenge will be achieving this without adding complexity. Financial terms have not been disclosed, nor the closing date, but the move has already put the industry on alert. The agentic era is here, and governance will be the differentiating factor between success and chaos.

As Michael Ni reflects, “Whoever governs agents, context, and autonomous actions will win the agentic era.” Snowflake is betting big, but time will tell if CIOs are ready to adopt these capabilities without stumbling along the way.

Original source: ComputerWorld. Analysis and adaptation by ForgeNEX.