SailPoint and Microsoft Ship AI-Agent Identity Controls as Fourth IAM Category Forms

Enterprise IAM Now Covers Four Identity Types, Not Three

SailPoint's October 6 launch of Agentic Fabric and Microsoft's October 12 Entra workload identity updates mark the formal separation of AI-agent identity from human and application identity. Enterprises that treat agent credentials as an API-management problem—provisioning static keys through existing application-onboarding workflows—will face audit gaps as agents operate across systems without human-equivalent access reviews. The shift requires procurement teams to answer four questions that existing IAM platforms do not address together: how agents are discovered, how credentials are issued, how runtime authorization is enforced, and how access is reviewed after use.

SailPoint positioned its announcement against this gap. Agentic Fabric and the redesigned Human Fabric share a governance model, allowing certification campaigns to cover employees and agents in a single workflow. The company's separation-of-duties engine evaluates access requests in real time and blocks combinations that violate policy before provisioning completes. The re-platformed certification engine removes record limits and data-delay constraints that previously forced enterprises to partition large access reviews into manual cycles. For on-premises customers, IdentityIQ 9.0 adds time-based access controls for roles and entitlements and refreshes the review interface without requiring a cloud migration.

Microsoft's approach integrates agent identity into existing Entra services rather than creating a separate product. Entra Agent ID provisions identities automatically for agents built in Copilot Studio. Continuous Access Evaluation now reacts to device deletion, disablement, and noncompliance signals, extending real-time revocation beyond user-risk events. SAP SuccessFactors provisioning replaced long-lived credentials with Entra-managed, short-lived tokens, and existing provisioning jobs can be upgraded without recreation. A guided migration workflow moves organizations from Entra Connect Sync to Entra Cloud Sync with built-in readiness checks and validation steps. Entra ID Governance discovers groups and memberships in connected applications, including orphaned groups and users without corresponding Entra assignments.

Three Architectures, No Convergence Yet

Microsoft, Okta, and ServiceNow have shipped agent identity products with incompatible architectures. Microsoft provisions agent identities within its application ecosystem. Okta's Agent SSO, launched September 22, replaces static API keys with identity-governed, short-lived tokens. ServiceNow's AI Passport, announced October 15, uses a cryptographically signed credential that agents check at runtime to verify authorized scope. SailPoint emphasizes governance and lifecycle management across agent types. These products address adjacent problems today but will compete for the control plane as agent counts grow and enterprises consolidate identity stacks.

RSA Agent ID, scheduled for general availability on November 16 for Discover and Secure modules and first half of 2027 for Govern, targets regulated industries. The staged release separates agent inventory and protection from certification and policy controls. Procurement teams in government, finance, and healthcare should plan discovery and security budgets for 2026 and governance budgets for 2027 separately.

What Buyers Must Verify Before Committing Budget

Neither SailPoint nor Microsoft disclosed subscription pricing, independent scale benchmarks, or named customer deployment figures. Buyers evaluating SailPoint should test the no-record-limit claim and real-time separation-of-duties enforcement with their own entitlement volumes and certification-cycle timings before signing contracts. Microsoft customers should verify which continuous-evaluation signals work across their actual client and resource mix, confirm licensing boundaries for workload identity features, and determine whether governance discovery covers their connected applications.

The agent identity gap exposes a procurement mistake: assuming that a vendor controlling one IAM layer—employee authentication, application provisioning, or access governance—automatically governs agents. Enterprises need to map four distinct capabilities and determine which vendors supply each. Agent discovery identifies where agents exist and what resources they touch. Credential issuance provisions identities and secrets. Runtime authorization enforces policy when agents act. Access review audits agent behavior after the fact. Buying only provisioning leaves runtime behavior ungoverned. Buying only governance leaves credential issuance uncontrolled. Enterprises that select a single vendor for agent identity should demand evidence that all four layers are covered, not just the one the vendor historically supplied for human users.

What to Watch

Agent identity spending will shift from API-management line items to IAM budgets over the next 12 months as audit teams demand certification campaigns that include non-human actors. Enterprises with separation-of-duties policies should prioritize platforms that enforce policy at request time rather than discovering violations during quarterly reviews. Organizations running hybrid directories should evaluate whether Microsoft's Entra Cloud Sync migration workflow reduces operational cost enough to justify consolidating around Entra or whether SailPoint's IdentityIQ 9.0 update extends on-premises deployment life. Procurement teams should separate agent discovery and security budgets from governance budgets if evaluating RSA or other vendors with staged release schedules.