AI touches identity from two directions. It makes an experienced consultant dramatically more productive — and it creates a fast-growing population of new identities that need to be secured. We work on both sides.
AI as a consultant’s force multiplier
AI doesn’t replace identity expertise — it amplifies it. We use AI tools deliberately throughout an engagement so senior judgment goes further, and you pay for outcomes rather than hours of legwork.
- Faster discovery and assessment — AI-assisted analysis of entitlement exports, configurations, and policies to surface outliers, dormant access, and role candidates in days instead of weeks.
- Legacy translation — decades of custom ISIM workflows, TDI assembly lines, and JavaScript provisioning rules mapped into clear, like-for-like requirements for Okta or SailPoint.
- Build work — scripts, Okta Workflows, Terraform, and API integrations drafted quickly, then reviewed and tested by an architect before delivery.
- Documentation and audit support — designs, runbooks, and policy-to-NIST control mappings produced alongside the work instead of left for the end.
How we use AI responsibly
- Strict Zero Trust data principles: client architecture and proprietary data never train public AI models, and are used only in tools and environments you approve.
- AI drafts; a senior architect reviews, tests, and is accountable for every deliverable.
- Nothing reaches your production environment without your review and approval.
Securing AI agents and MCP
Most organizations have no inventory, lifecycle, or authorization model for their AI agents. The Model Context Protocol (MCP) makes it easy to connect agents to tools and data — and just as easy to connect them to everything.
- MCP server authorization — OAuth 2.1-based authorization for remote MCP servers, protected-resource and authorization-server metadata, client registration strategy, and scoped tokens per tool.
- Your IdP as the authorization layer — put Okta or Auth0 in front of agent tool access rather than scattering API keys and static secrets.
- Gateway patterns — centralize MCP traffic through an identity-aware proxy or API gateway for policy enforcement, logging, and kill-switch control.
- On-behalf-of delegation — token exchange so an agent acts with the user’s permissions, never more; evaluation of emerging standards such as Okta Cross App Access (XAA).
- Fine-grained authorization for RAG — relationship-based access control with OpenFGA / Okta FGA so retrieval only returns what the requesting user is allowed to see.
Governing non-human identities
- Inventory agents, service accounts, API tokens, and MCP clients.
- Bring them under lifecycle and certification in Okta Identity Governance or SailPoint — owners, expiry, and periodic review.
- Detect over-privileged and orphaned credentials before an agent finds them.