Immuta Introduces the First Data Provisioning Platform for Managing Agentic Data Access

New platform capabilities enable enterprises to provision and govern AI agent access to enterprise data in real time — without impersonation, standing privileges, or ticket-driven delays

Immuta, the Data Provisioning Company, introduced the first data provisioning platform for managing agentic data access. The launch includes new Agentic Data Access capabilities that enable enterprises to provision and govern AI agent access to enterprise data in real time. With Agentic Data Access, Immuta treats AI agents as first-class identities — with their own attributes, intent, temporary access, and audit trail — so they can act on behalf of users without authenticating as them, standing privileges, or ticket-driven delays.

The launch addresses a major shift in enterprise AI: agents are becoming a primary interface for interacting with enterprise data, and they need to be governed not as shadow users or generic service accounts, but as first-class actors in the data ecosystem.

AI agents are a new class of data consumer. They operate 24/7, generate queries dynamically, move across systems continuously, and expect access decisions in seconds, not days. As organizations move from AI experimentation to enterprise deployment, they need to prepare for a future where non-human identities may outnumber human users requesting access to data.

Most data provisioning models were never built for that world. Traditional approaches rely on static roles, manual approvals, and ticket-driven workflows — often routed through systems such as ServiceNow or Jira — that were designed for infrequent and predictable requests from people. In many early AI deployments, the default approach has been to let the agent log in as the user asking the question. That may be enough for a pilot. It does not work at enterprise scale.

At enterprise scale, that model breaks down quickly. It drives account sprawl because every potential AI user must be provisioned across every system an agent might touch. It increases exposure risk because teams grant broader, longer-lasting permissions to prevent agents from failing mid-task. It leads to rights inflation, especially when agents inherit the privileges of powerful users whose access far exceeds what the task requires. And it creates a muddy audit trail, because logs may show a human executed a query when the action was actually performed by an agent. In practice, it creates human-limited AI: agents stall when they hit semantic or permission walls, and the business gets less value from the systems it is trying to scale.

Immuta’s Agentic Data Access capabilities are designed to solve that problem at the authorization layer. When an agent needs data, Immuta evaluates the request against centrally defined policies based on who the agent is acting for, what data is being requested, and why it is needed. Immuta then provisions temporary access directly in the underlying data platform — such as Snowflake, Databricks, and BigQuery — granting only the access required for that task and automatically removing it when the task is complete.

Marketing Technology News: MarTech Interview with Liat Barer, Chief Product Officer @ Odeeo

How Immuta’s Agentic Data Access Works

Immuta’s Agentic Data Access capabilities extend its policy-driven data provisioning model to AI agents, enabling enterprises to govern non-human identities with the same precision as human users — and with the speed agent-driven workflows demand.

• Agents as First-Class Actors: Immuta treats AI agents as first-class actors in the data ecosystem, with their own identity, attributes, intent, and audit trail. Rather than operating through OAuth or shared credentials, agents are governed explicitly and transparently.
• Role Vending for AI Agents: When an AI agent acts on behalf of a user, Immuta dynamically generates a temporary role in the underlying data platform that reflects only the permissions the user is authorized to exercise for that specific task. This is deterministic policy enforcement at the data platform tier, unlike non-deterministic solutions at the LLM tier.
• Zero Standing Privileges: Access is provisioned just in time and removed automatically when the task is complete, reducing persistent exposure, eliminating unused credentials, and minimizing risk.
• Clear, Auditable Activity: Immuta records when an AI agent accessed data, which user it acted for, and what data was retrieved, creating a clear audit trail for governance, security, and compliance teams.

This gives enterprises a fundamentally different model for the age of agents. Instead of collapsing the agent and the user into one identity, Immuta gives them separate identities that work together. The agent acts on behalf of the user, not as the user. That reduces account sprawl, eliminates standing privileges, and creates clear accountability for when a person accessed data, when an agent acted independently, and when an agent acted on behalf of a user.

“AI agents are not just another application endpoint — they are a new identity type in the enterprise,” said Matthew Carroll, CEO of Immuta. “They are always on, they move at machine speed, and they need access in the moment. If companies keep trying to provision them through workflows built for humans, they will either bottleneck AI or overexpose sensitive data. Agentic Data Access gives them a way to move fast without giving up governance.”

Marketing Technology News: What Marketers Need to Know About the European Accessibility Act

The Next Era of Data Provisioning

This announcement marks the next phase in Immuta’s evolution — from separating policy from platform, to automating data provisioning, to now governing agent-driven access at machine speed. Because Immuta externalizes policy and enforces it directly within cloud data platforms, extending that model to AI agents is a natural evolution of the platform, not a reinvention.

What’s Next: From Access to Understanding

Immuta’s Agentic Data Access capabilities are the foundation for a broader shift in how data is discovered, accessed, and used. Granting the right access at the right moment is only the beginning. The next step is helping agents understand what data a user is allowed to use, how that data is defined, and how to safely request more access when needed.

Looking ahead, Immuta is extending this model in two key areas:

• Semantic Governance: Immuta will extend access awareness into the semantic layer, giving agents visibility into which metrics, fields, and relationships a user is allowed to use — and for which purposes — while enforcement continues in the underlying data platform. This helps agents plan better queries, avoid dead ends, and generate more accurate results.
• Agent-Initiated Access Requests (Question-Time Access): When additional data is required to answer a question, agents will be able to request access on behalf of users in real time. Requests can be automatically approved through deterministic policies or routed for human review when additional oversight is needed, enabling a new model of just-in-time, question-driven access provisioning.

Together, these capabilities move beyond access control to access orchestration, enabling agents to safely discover, request, and use data in a fully governed, fully auditable way at machine speed.

“We are moving into a world where more data access requests will come from agents than from people,” said Steve Touw, CTO of Immuta. “That changes the fundamentals of provisioning. The organizations that win will be the ones that can grant the right access for the right task at the right time, while preserving governance, accountability, and control. That is the future Immuta is building for.”

As AI agents become embedded across analytics, operations, and decision-making workflows, enterprises will need a new provisioning model built for both human and non-human identities. With Agentic Data Access, Immuta is delivering the foundation for that next era — and the roadmap for what comes after.