Skip to content
Last updated on

AI agents

AI agents are autonomous software programs that perceive their environment, make decisions, and take actions to achieve specific goals. Okta for AI Agents lets you register, configure, and manage AI agents in your org so that they operate with least privilege and remain accountable members of your digital workforce.

Overview

Okta for AI Agents manages the registration, configuration, and lifecycle of AI agents, helping to ensure the following:

  • AI agents are registered and validated before connecting to resources
  • Apps are linked to AI agents for human-access accountability
  • Public key credentials are configured correctly for agent authentication
  • Access is authorized through standardized token exchange flows

Relationships

AI agent token request relationships

linked for human access to workloads

request ID-JAG tokens from

may exchange ID-JAG for access token

for authentication

for authorization

protecting a

AI-AGENT

APPLICATION

ORG-AUTHORIZATION-SERVER

CUSTOM-AUTHORIZATION-SERVER

JSON-WEB-KEY

CONNECTION

PROTECTED-RESOURCE

linked for human access to workloads

request ID-JAG tokens from

may exchange ID-JAG for access token

for authentication

for authorization

protecting a

AI-AGENT

APPLICATION

ORG-AUTHORIZATION-SERVER

CUSTOM-AUTHORIZATION-SERVER

JSON-WEB-KEY

CONNECTION

PROTECTED-RESOURCE

Related developer documentation:

AI agent user sign-on app

is linked to an

on behalf of a

token exchanges

AI-AGENT

APPLICATION

USER

ID-TOKEN

is linked to an

on behalf of a

token exchanges

AI-AGENT

APPLICATION

USER

ID-TOKEN

Connection

authorizes

to use

with an

for retrieving a

Connection

string

type

Principal

Protocol

AuthorizationServer

Token

authorizes

to use

with an

for retrieving a

Connection

string

type

Principal

Protocol

AuthorizationServer

Token

Okta for AI Agents lifecycles

Okta for AI Agents objects have lifecycles that are driven by system and user interactions.

When using these APIs, it's important to understand the potential status values of objects, and how they transition from one lifecycle state to another.

AI agent status lifecycle

POST ../ai-agents

system process

POST ../activate

POST ../deactivate

DELETE ../ai-agents/{agentId}

not-created

operation-started

STAGED

ACTIVE

INACTIVE

404 Not Found

POST ../ai-agents

system process

POST ../activate

POST ../deactivate

DELETE ../ai-agents/{agentId}

not-created

operation-started

STAGED

ACTIVE

INACTIVE

404 Not Found

AI agent status descriptions

  • OPERATION-STARTED: An AI agent creation request has been submitted. Okta is processing the registration asynchronously.
  • STAGED: The AI agent has been created and is ready to be activated. Initial state after successful registration.
  • ACTIVE: The AI agent is active and can authenticate to authorization servers and connect to resources.
  • INACTIVE: The AI agent is deactivated and can't authenticate or connect to resources.
  • DELETED: The AI agent has been deleted and removed from the system.