Task Budget

Simple Definition

A task budget is the expected tokens, cost, time, model calls, tool calls, retries, and human review effort a task will need, estimated before the work starts rather than discovered after the invoice arrives.

Why It Matters

Agent work isn’t one predictable prompt, it’s a chain of steps that can grow if the agent hits friction. Setting a task budget up front gives you a low, expected, and high range to plan against, and a trigger point: if actual usage is tracking well past the high estimate, that’s a signal to check in rather than let the task run unchecked.

What Goes Into a Task Budget

  • Context and input size
  • Expected output size
  • Likely number of tool calls
  • Expected agent steps and realistic retries
  • Human review time

Example

Before assigning a coding agent to fix a bug, a developer budgets: moderate context (the relevant files), a couple of tool calls (search, run tests), one likely retry if the first attempt doesn’t pass, and a real code review before merging. If the agent blows well past that budget without a clear reason, that’s the cue to step in.

For the full estimation framework, see How to Estimate an AI Agent’s Cost.

Continue learning

Explore related guides, tools, workflows, and prompts that help you go deeper into this topic.

See AI terms in action

Browse practical AI workflows that use the concepts in this glossary.

Last updated: