Why it matters
An agent may interact with models, tools, and business applications during a task. Those actions create a need for clear identity, boundaries, and human review.
How it works
An agent can use a model to interpret a goal, select an available tool, and continue through a configured workflow. The level of autonomy depends on its design and permissions.
How Halo approaches it
Halo Agents combine model choice, app connections, policy, approvals, and budgets. Halo gateways and the secure agent runtime apply controls to supported agent traffic and actions.