Skip to main content

Task Execution Autonomy (TEA)

The first axis of the Autonomy Spectrum Framework — the proportion of a business's revenue-generating task volume executed by agents rather than people, scored 0–2 against who moves the work through the Revenue Loop.

Extended Definition​

Task Execution Autonomy is the first axis of the Autonomy Spectrum Framework: the proportion of a business's revenue-generating task volume executed by agents rather than people, scored 0–2 against who moves the work through the Revenue Loop. The axis measures who executes, not who decides. It asks one question of every unit of work in the loop: when the work moved, did a person move it or did the system?

TEA is deliberately blind to decision-making, which belongs to Decision Execution Autonomy, and to workflow continuity, which belongs to the Process Continuity Score. It isolates the execution layer because execution is where most autonomy claims are made and where most of them fail inspection. An agent executes a task only when no person touches the path between the trigger and the completed output; a person who initiates, completes, or reviews-and-releases a task has executed it, however much the technology assisted.

A score of 0 indicates that people execute the revenue-generating workflow, with technology assisting them — drafting, suggesting, accelerating, but never owning the task. A score of 1 indicates that agents own material task volume within the Revenue Loop, but people still initiate, complete, or carry critical segments of it. A score of 2 indicates that agents own the revenue-generating task volume end to end, with people absent from the execution path and present only in the Judgment Layer.

The axis exposes the most common misreading of AI adoption. A business in which every employee uses AI tools daily, and in which those tools have doubled individual productivity, scores 0 — because the people are still executing. Tool adoption changes the speed of human execution; it does not change who executes. TEA registers only the transfer of execution itself.

Application​

In Autonomy Spectrum scoring, TEA is assessed against the T1 and T2 composition of the Revenue Loop: the share of routine and conditional task volume owned end-to-end by agents, measured at the loop level rather than the tool level. A business is scored on what its agents execute without a person in the path, not on which AI products its people use.

  • Revenue Loop — The Revenue Loop is the unit of analysis for TEA scoring: the axis measures the proportion of task volume within the loop that agents execute end-to-end without human initiation, completion, or review.
  • Task Tiers (T1 / T2 / T3) — Task Tiers determine which task volume TEA scoring targets: T1 and T2 tasks are the primary candidates for agentic execution; T3 tasks with mandatory human judgment cannot contribute to a high TEA score.
  • The 80 Percent Threshold — The 80 Percent Threshold is the operational benchmark that a score of 2 on TEA approaches: agents owning the revenue-generating task volume end-to-end is the execution expression of crossing that threshold.
  • Judgment Layer / Execution Layer — TEA scores the Execution Layer directly: a score of 2 means the Execution Layer is owned entirely by agents, with humans present only in the Judgment Layer for the exception classes the architecture deliberately assigns to the Steward.
  • Autonomy Spectrum Framework — Task Execution Autonomy is the first axis of the Autonomy Spectrum Framework, and the axis where most autonomy claims are made and where most of them fail inspection when scored against who actually moves the work.

References​

  • Lexicon — canonical definition
  • Wiki — extended entry

Metadata​

First used: 2026-06-12
Pillar: How We Think

Part of the Arco Lexicon Ecosystem — maintained by Arco Venture Studio