Isabelle Williams on safe agent access

AI Engineer18m 33s
Isabelle Williams in a lavender T-shirt beside the blue and white headline Safe Agent Access on black.
1 VIEW
0 comments · 0 votesOpen discussionClose discussion
Sign in to join the discussion

    Isabelle Williams uses a travel-planning application to explain how an agent can discover account and API access instructions. She describes an anonymous trial followed by a user claim flow that connects the work to an account.

    Williams demonstrates an identity-assertion flow in which the agent asks permission to share user details and the application decides whether to create an account and what access to grant. She emphasizes signed assertions, audience checks and the application’s control over trust.

    For an agent that can move travel funds, Williams issues a separate agent token with permissions limited by both the user and an agent-specific ceilingAn agent permission boundary defines and enforces which resources and actions an automated agent is allowed to use.. Her demonstration illustrates auditing, revocation and the difference between an agent changing a budget and a user changing it.

    Original YouTube thumbnailWatch on YouTube

    Video topic coverage

    • Agents & workflows100
    • Coding & tools70
    • Models & learning0
    • Infrastructure & compute57
    • Research & evaluation0
    • Safety & security98
    • Business & productivity27
    • Creative & media0
    • Robotics & perception0
    • Society & policy0

    Share this video

    Isabelle Williams in a lavender T-shirt beside the blue and white headline Safe Agent Access on black. With WWW.ARTIFICIAL-INTELLIGENCE.VIDEO, 23 October 2026 and duration 18m 33s.

    Isabelle Williams demonstrates how agents can discover an application, act with user consent and operate under limited, revocable permissions.