Legacy IGA vs Modern IGA: What Changed and Why It Matters
Modern IGA is identity governance delivered as a cloud service that evaluates access continuously, connects to applications through lightweight connectors, and governs people, service accounts, and AI agents in one model. Legacy IGA is typically self-hosted, customized through code, and organized around scheduled certification campaigns. The goals are the same. The operating model changed.
See what modern IGA looks like in practice →
Short answer: what changed#
Legacy identity governance and administration (IGA) platforms were built for a world where most applications ran on premises, most identities were employees, and auditors were satisfied by a quarterly certification campaign. Those platforms did the job, and many still do. Three things changed around them:
- Where applications live. Most enterprises now run hundreds of SaaS applications alongside the on-prem systems they still depend on.
- How fast access changes. Access decisions now happen in seconds, through automated workflows and AI agents, not in a quarterly review window.
- Who needs governing. Service accounts, API keys, and AI agents now hold real access, often more of it than people do.
The C1.ai 2026 Future of Identity Report, a survey of 508 IT and security leaders at U.S. organizations with more than 1,000 employees, shows the strain:
- 55% name system complexity as their top IAM challenge, followed by tooling limitations at 41%.
- 41% still rely on standing permissions, while only 19% use continuous policy-based enforcement.
- 27% say their existing identity tools were not designed for AI agents.
Modern IGA is the response to those three shifts. It keeps the same governance fundamentals and changes how they run.
Legacy IGA vs modern IGA at a glance#
The right-hand column is what to ask any vendor, legacy or modern, to prove in a proof of concept.
| Legacy IGA | Modern IGA | How to check it | |
|---|---|---|---|
| Deployment | Self-hosted on customer infrastructure: application server, database, and the staff to run them. Upgrades are projects with regression testing. | SaaS run by the vendor. Updates ship continuously with no customer downtime. | Ask how many upgrade projects the last three releases required, and who did the work. |
| Connectors | Connectors configured inside the platform, often with custom correlation and provisioning rules per application. | Prebuilt cloud connectors, an on-prem agent for systems behind the firewall, and file, SQL, or HTTP connectors for apps with no API. | Bring your list of in-scope apps and count how many connect without custom code. |
| Time to onboard an app | Often scoped as a project per application. | Connect, sync, and review in the same platform. | Time one real app from connection to first review in production. |
| Access reviews | Scheduled campaigns, quarterly or annually, built by an administrator. | Scheduled campaigns plus event-driven reviews, usage context on every line, and automatic revocation. | Measure review cycle length and reviewer completion rate. |
| AI agents and NHIs | Built around people with HR records. Non-human identities are often tracked somewhere else. | People, service accounts, and AI agents in one inventory, each with a human owner. | Ask to see an AI agent's owner, entitlements, and last review in the product. |
| Admin effort | Customization through Java, BeanShell, or XML rules. Often depends on one or two specialists. | Policy configuration, APIs, and infrastructure as code. | Ask who on your team could change an approval workflow today. |
| Cost | License plus infrastructure, professional services, and specialist staff time. | Subscription, with lower services and infrastructure overhead. | Compare total cost of ownership, including staff hours, not just license. |
Continuous governance vs quarterly campaigns#
The biggest change is not where the software runs. It is how often it makes decisions.
Legacy IGA treats governance as an event. Access is granted through a request, then checked during a campaign a few months later. Between campaigns, access drifts: people change teams, projects end, contractors leave, and entitlements pile up. The campaign catches some of it, usually in a rush before the audit.
Continuous identity governance treats governance as a running process. Access is evaluated when something changes, not only when the calendar says so:
- Lifecycle events drive access. When HR records a hire, move, or departure, joiner-mover-leaver policies grant, adjust, or remove access the same day.
- Policy decides routine requests. Low-risk requests that meet policy are approved automatically. High-risk requests go to the right approver.
- Access expires. Time-bound and just-in-time grants end on their own instead of waiting for a reviewer to find them.
- Reviews are triggered by risk. A role change, a new admin grant, or an unused entitlement can start a focused review outside the campaign cycle.
- Conflicts are caught early. Separation of duties checks run before access is granted and keep running after.
This matches the direction of NIST's zero trust guidance (SP 800-207), which calls for access to be decided by dynamic policy rather than fixed, long-lived grants.
AI agents make the case sharper. The Future of Identity Report found that 64% of organizations allow AI agents to act autonomously within policies or with post-action review. A quarterly campaign cannot review a decision an agent made in seconds. Policy has to do that work in real time, and the campaign becomes the audit record of how well policy did it.
Campaigns do not go away. Auditors still expect periodic certification, and PCI DSS v4.0 requirement 7.2.4, for example, calls for a review of user accounts and access privileges at least every six months. What changes is how much work each campaign takes, because continuous governance has already removed most of the stale access before reviewers see it.
Where modern IGA changes the numbers#
"Modern" is easy to claim. These are the numbers that show whether it is true, with examples from C1.ai customers.
| Metric | How to measure it | C1.ai customer example |
|---|---|---|
| Access review cycle | Days from campaign launch to sign-off | PriceSmart cut its quarterly access review cycle from three months to two to three weeks. |
| Reviewer completion rate | Share of assigned reviews completed by the deadline | DigitalOcean completed 1,200 access reviews across seven departments with 85% less effort and a 100% reviewer completion rate. |
| Access request effort | IT hours spent processing requests | Ramp reduced the IT effort required to process access requests by 95%. |
| New hire provisioning | Time from start date to full access | Zscaler cut new hire provisioning from weeks to 10 minutes. |
| Review preparation | Time to scope and launch a campaign | A large pharmaceutical company reduced access review preparation from three weeks to under three days. |
| Time to onboard an app | Days from connection to first review in production | Built onboarded 200+ apps in two days. |
Pick three of these before a proof of concept, record your current baseline, and measure the same three at the end. That turns "modern" from a label into a result.
AI agents and non-human identities#
Legacy IGA assumes an identity is a person with a manager, a department, and an HR record. That assumption no longer covers most of the access in an enterprise. According to the Future of Identity Report:
- 95% of organizations have AI agents performing at least one IT or security task autonomously.
- 47% say non-human identities outnumber human users.
- Only 22% report full visibility into their non-human identities, down from 30% the year before.
- 45% already use IAM tools for NHI governance, and another 45% plan to within 12 months.
Modern IGA governs agents with the same controls it uses for people. Every agent has an owner who is accountable for it. Its entitlements are scoped to the task and reviewed alongside human access. When the owner leaves or the project ends, its access is revoked. One inventory, one review process, and one audit trail, instead of a separate blind spot for every new agent. For more on this, see What is AI agent identity?
What stays the same#
Modernizing IGA does not mean starting governance over. The fundamentals carry forward:
- The controls. Least privilege, separation of duties, joiner-mover-leaver, and periodic certification are still the core of the program.
- What auditors ask for. SOX, HIPAA, PCI DSS, and ISO 27001 still expect evidence that access was approved, appropriate, and reviewed. Modern platforms produce that evidence faster. They do not change the requirement.
- The need for clean data. An authoritative identity source, usually the HR system, still drives everything. Bad source data produces bad decisions on any platform.
- Role design. Roles still matter. Modern platforms make them easier to build from real access through role mining, but someone still has to own them.
- The people. Application owners, managers, and security teams still make the judgment calls. The goal is fewer, better decisions, not zero decisions.
Modernization also does not have to happen all at once. A common path is to move access reviews to a modern platform first while the existing system keeps provisioning, then migrate provisioning and lifecycle in waves. Teams running SailPoint can see how that works in our guide to SailPoint IdentityIQ migration.
Signs it is time to modernize#
None of these mean your current platform has failed. They mean the cost of staying put is rising:
- A major upgrade or platform migration is on the calendar, and the switching cost is being paid anyway.
- One or two specialists are the only people who can change a workflow.
- Quarterly access reviews take most of the quarter.
- Applications in audit scope sit outside governance because connecting them is a project.
- AI agents and service accounts hold access with no named owner.
For a full plan, start with Identity Modernization: Moving from Legacy to Modern IGA, and for the fundamentals see What is IGA?
FAQ#
What is modern IGA?#
Modern IGA is identity governance and administration delivered as a cloud service. It evaluates access continuously through policy, connects to cloud and on-prem applications through lightweight connectors, and governs human and non-human identities in one model.
What is the difference between legacy IGA and modern IGA?#
Legacy IGA is usually self-hosted, customized through code, and built around scheduled certification campaigns for human users. Modern IGA is SaaS, configured through policy and APIs, and governs access continuously for people, service accounts, and AI agents.
What is continuous identity governance?#
Continuous identity governance evaluates access whenever something changes, such as a hire, a role change, or a new grant, instead of only during periodic review campaigns. It relies on lifecycle automation, policy-based approvals, time-bound access, and event-driven reviews.
Is traditional IGA still effective?#
Traditional IGA still runs certifications and provisioning for many enterprises. It is harder to scale to hundreds of SaaS applications and to non-human identities, and changes often depend on specialist staff.
Do I have to replace my legacy IGA all at once?#
No. Many teams move access reviews to a modern platform first while the legacy system keeps provisioning, then migrate provisioning and lifecycle in waves.
How does modern IGA govern AI agents?#
Modern IGA treats each AI agent as an identity with a human owner, scoped entitlements, inclusion in access reviews, and automatic revocation when the owner leaves or the work ends.
How does C1.ai approach modern IGA?#
C1.ai is a cloud IGA platform that governs people, service accounts, and AI agents in one model. It connects to SaaS and on-prem applications, including systems with no API through file connectors, automates joiner-mover-leaver and access requests through policy, and runs access reviews with usage context and automatic revocation.
Book a demo → · See modern IGA in practice →
Sources: C1.ai 2026 Future of Identity Report (survey of 508 IT and security leaders at U.S. organizations with 1,000+ employees) · NIST SP 800-207, Zero Trust Architecture · PCI DSS v4.0, requirement 7.2.4 · C1.ai customer case studies: PriceSmart, DigitalOcean, Ramp


