Autonomous Agent & Automation Terms
CoreForge Atlas may allow users to create, configure, supervise, or operate specialized AI agents.
User Authorization
An Atlas agent may act only within permissions granted through the applicable Service and connected systems.
Users are responsible for determining what access agents should receive.
Connected Accounts
Atlas may connect to third-party platforms.
Users represent that they are authorized to connect each account and authorize requested actions.
Agent Actions
Depending on configuration, agents may perform actions such as:
• drafting communications; • organizing information; • analyzing documents; • creating tasks; • interacting with connected software; • processing business information; or • performing other authorized workflows.
High-Risk Actions
Creator CoreForge may require human confirmation before designated consequential actions.
Product architecture should provide configurable approval gates for activities such as:
• spending money; • executing contracts; • permanently deleting information; • changing critical permissions; • publishing public content; • sending sensitive communications; • initiating financial transactions; and • other high-impact activities.
AI Errors
Agents may misunderstand instructions, choose incorrect tools, hallucinate information, duplicate actions, or perform unintended actions.
Users must supervise agents according to the importance and risk of the task.
Permissions
Creator CoreForge should implement least-privilege authorization wherever technically feasible.
Logging
Important agent activity should be logged to provide accountability, debugging, security, and user visibility.
Revocation
Users should be able to revoke agent permissions and disconnect supported third-party integrations.
Enterprise Data
Where Atlas processes information on behalf of business customers, additional data-processing and enterprise agreements may apply.