Does using AI tools count toward Task Execution Autonomy (TEA)?

No. If a person initiates, completes, or reviews-and-releases the task, the execution is human regardless of how much the tools assisted. TEA registers only the transfer of execution itself — work where no person touches the path between trigger and completed output. Tool adoption increases the speed of human execution; it does not change who executes, which is the only variable TEA measures.

Why is TEA scored only against revenue-generating work?

Because the Revenue Loop is where execution ownership has structural consequences for cost structure and scaling. A business that automates internal reporting while humans deliver the product has not changed who executes its operation — it has changed the paperwork. TEA focuses on the Revenue Loop because that is where a score of 2 produces a genuinely different kind of business, not just a faster version of the same one.

Can a business score 2 on TEA and still not be autonomous?

Yes. TEA measures execution only. A business where agents execute everything but humans make every routing decision will score 2 on TEA and 0 on Decision Execution Autonomy (DEA). A business where agents execute within each stage but humans gate every handoff will score 2 on TEA and 0 on Process Continuity Score (PCS). A business where agents handle all execution but the Steward intervenes hourly will score 2 on TEA and 0 on Intervention Dependency (ID). The five-axis framework exists precisely because a high score on any single axis is insufficient — autonomy requires all five.

What is the relationship between Task Execution Autonomy (TEA) and the Human-to-Logic Ratio?

The Human-to-Logic Ratio (HLR) is Arco's market selection metric — the proportion of a business's gross margin consumed by human labour costs. TEA is the operational measurement of whether that ratio is being reduced through genuine execution transfer. A market with a high HLR presents a large opportunity for autonomous reconstruction; a business deployed in that market scores on TEA by how much of that human labour it has moved to agents. A business can operate in a high-HLR market and still score 0 on TEA if it has adopted tools without transferring execution. HLR identifies the opportunity; TEA measures whether the opportunity is being realised.

How does TEA relate to the other four axes of the Autonomy Spectrum Framework?

TEA answers who moves the work. The other four axes answer related but distinct questions that TEA cannot. Decision Execution Autonomy (DEA) asks who decides how that work is routed and resolved — a business can score 2 on TEA while every routing decision is made by a person. Process Continuity Score (PCS) asks whether the work flows without halting between stages — a business can have full agent execution within each stage while humans gate every handoff. Intervention Dependency (ID) asks how long the system runs before it needs a person — a business can have high TEA while a Steward intervenes hourly. Structural Headcount Independence (SHI) asks whether growth requires hiring — a business can transfer execution to agents while its growth model still demands proportional headcount. None of the four can be inferred from any of the others, which is why the composite score requires all five.

Who Moves the Work

When a unit of work moves through a business, either a person moved it or the system did. There is no third answer. Every invoice issued, every ticket resolved, every document processed, every order fulfilled was carried by a human hand or by encoded logic — and the proportion between the two is the most basic fact about how a business actually operates.

Task Execution Autonomy (TEA) 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. It is the axis where most autonomy claims are made, and the axis where most of them fail inspection.

Execution, not assistance

The axis draws a line that the market has worked hard to blur. A person who drafts an email with a model's help executed that task. A person who reviews and sends an agent's draft executed that task. A person who approves a payment an agent prepared executed that task. In each case the technology assisted, accelerated, perhaps transformed the work — and in each case a human remained in the execution path, which means the execution was human.

An agent executes a task when no person touches the path between the trigger and the completed output. The document arrives, is classified, processed, validated, and filed — and the first human contact with that unit of work, if there is one, is reading about it in a digest. That is the standard the axis scores against, and it is deliberately unforgiving, because the alternative standard — counting assisted work as autonomous work — makes every business with a software subscription autonomous.

The scoring semantics

A score of 0 indicates that people execute the revenue-generating workflow, with technology assisting them. 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 Layer and governing only from the Judgment Layer / Execution Layer — the exception classes the architecture has deliberately assigned to the Steward. A score of 2 on TEA approaches what the framework calls The 80 Percent Threshold: the operational benchmark at which more than 80% of cross-departmental handoffs execute without human intervention.

Two boundaries in those semantics deserve attention. The first is the phrase revenue-generating. TEA is scored against the Revenue Loop, not against the whole of a company's activity, because the Revenue Loop is where execution ownership has structural consequences. A business that has automated its internal reporting while humans deliver the product has automated its paperwork, not its operation.

The second is the boundary between a score of 1 and a score of 2. The difference is not volume — it is the critical path. A business where agents handle ninety percent of task volume while humans carry the segment every transaction must pass through has built an assistance layer around a human bottleneck. The bottleneck owns the loop.

Why tool adoption scores zero

This is the finding that makes TEA uncomfortable, and therefore useful. A business in which every employee uses AI tools daily, in which individual productivity has doubled, in which the leadership describes the company as AI-first in every investor update — that business scores 0 on the first axis, because the people are still executing. The tools changed the speed of human execution. They did not change who executes.

The distinction matters economically, not just semantically. A business whose execution is human scales its execution by hiring, whatever its tools cost. A business whose execution is agentic scales through Labor-to-Compute Substitution — provisioning compute rather than people. These are structurally different companies with different cost structures, different margin curves, and different scores on Structural Headcount Independence (SHI), the fifth axis that quantifies exactly this consequence at scale. A measurement framework that scored them identically would be measuring nothing. The market's confusion between adoption and autonomy is not a small category error. It is the category error, and TEA exists to make it visible as a number.

What the axis does not measure

TEA is deliberately blind to everything but execution. It does not ask who decided how the work was routed — that is Decision Execution Autonomy (DEA). It does not ask whether the workflow halts between stages — that is the Process Continuity Score (PCS). A business can score 2 on TEA while a human approves every transition and intervenes every hour; the other axes exist to catch exactly that. The first axis answers one question precisely rather than five questions vaguely: when the work moved, who moved it?

The Operator's Verdict

Ask the question of your own operation across each Task Tier (T1, T2, T3) and count honestly. Every task where a person initiates, completes, reviews-and-releases, or carries the handoff is a human execution, whatever assisted it. Most operators who run this count discover that their AI transformation has transferred almost no execution at all — which is not a failure of the tools. It is the difference between buying technology and re-architecting a business.

Technology changes what is possible. Execution determines what is transferred.

KEY TAKEAWAY

What is Task Execution Autonomy (TEA)?

Task Execution Autonomy (TEA) 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. A score of 0 means people execute with technology assisting; a score of 1 means agents own material task volume but people still carry critical segments; a score of 2 means agents own the revenue-generating work end to end, with people present only in the Judgment Layer. A business where every employee uses AI tools daily still scores 0 if the people are executing, because tool adoption changes the speed of human execution through Labor-to-Compute Substitution — not who executes. TEA is scored against the Revenue Loop only, because the Revenue Loop is where execution ownership has structural consequences for cost structure and scaling. The axis is deliberately blind to routing decisions (Decision Execution Autonomy), workflow continuity (Process Continuity Score), intervention frequency (Intervention Dependency), and headcount scaling (Structural Headcount Independence) — each of which has its own axis.