Sample report

See what an AgencyThread report actually looks like

This report is built from a synthetic market-entry scenario created specifically to demonstrate the product. Aster Vale Coffee is fictional, its market data are synthetic, and the workflow was created to demonstrate AgencyThread rather than drawn from anyone's private work.

The frozen synthetic case was run through AgencyThread’s canonical evaluation process before any result was known, so you can see how AgencyThread interprets:

Human Agency Index;
Collaboration Profile;
Evidence Confidence;
what strengthened the result;
what limited the result;
and how the human and AI worked together.

Synthetic demonstration

The evidence was frozen before evaluation, and no target Human Agency Index or Collaboration Profile was specified in advance. This demonstration does not establish real-world or external validity.

Synthetic Demonstration Case

Aster Vale Coffee — U.S. Market Entry RecommendationSynthetic demonstration · Strategy-manager-led market selection

The evidence supports High Human Agency in this market-entry recommendation. The strategy manager framed the decision as a learning-market choice before involving AI, rejected the AI's first analytical model on substantive grounds, originated the staged flagship-plus-conditional-second-store entry design that became the centerpiece of the final recommendation, and withheld final adoption until evidence limitations and capital math were checked. AI in turn substantively rebuilt models, ran multi-scenario sensitivity analysis, surfaced factors the first pass had missed, and developed a genuine counter-case for the market that was not chosen. This was a reciprocal, evolving exchange rather than a one-shot request-and-approval process, which is why the evidence also supports a Co-Intelligence collaboration pattern.

Human Agency Index90High Human Agency

Collaboration Profile ExperimentalCo-Intelligence3 of 3 evaluations

Evidence ConfidenceModerateResult consistency: Strong

Low0–39Moderate40–59Substantial60–79High80–100*

*High is subject to the locked High-HAI gate.

Collaboration Profile Experimental

Human-Led Creation

AI-Augmented Human Work

Co-Intelligence

AI-Led Collaboration

Delegated Generation

High Human Agency and Co-Intelligence are not in tension here: the manager remained intellectually consequential throughout the process while AI performed a large share of the substantive modeling, scenario development, and drafting.

Why this result

How the manager and AI actually worked

The human’s role

The strategy manager reframed the assignment from “best coffee city” to “best first U.S. learning environment” before any AI involvement, then repeatedly redirected the analysis on substantive grounds: rejecting a visibility-heavy first model, requiring multi-scenario sensitivity analysis instead of a single score, catching a concept-test sample-size error, and demanding a good-faith counter-case for the market that was not chosen. The manager originated the staged, gated entry design the final recommendation is built around, and made the final call on judgment about what the first market needed to teach the company rather than on which option scored highest.

AI’s role

AI built and rebuilt the analytical weighting models, ran the four-scenario sensitivity analysis, surfaced factors the initial pass had not captured, and developed the staged pilot concept into a detailed operating plan. It also produced a genuine, substantive counter-case for the rejected market rather than a token rebuttal, and drafted the final recommendation document. This was a materially AI-developed process, not a lightly AI-assisted one: AI's analysis repeatedly changed what the human considered next.

What strengthened the result

The strongest evidence came from problem framing, critical evaluation, and constructive development. The manager's pre-AI reframing organized everything that followed; the record shows repeated, well-reasoned acceptance, rejection, and qualification of AI output rather than either blanket approval or reflexive disagreement; and the staged two-format pilot design was a genuine human-originated solution mechanism, not a governance or approval action credited to development.

What limited the result

Workflow direction and monitoring were the relatively less strongly evidenced dimension, though still solidly supported: the task decomposition emerged turn-by-turn in response to intermediate results rather than as a single pre-declared multi-stage plan, and while the manager caught a real data error and scoped verification appropriately, the record shows fewer instances of anticipatory risk detection before an issue had already surfaced. This is a matter of degree within an already strong record, not a weak dimension.

Human Agency Profile

Six dimensions, one continuum

Direction & Control

D1Framing & Ownership

Problem Framing & Evolving Ownership

The evidence strongly supports the manager's problem framing and evolving ownership: the learning-market reframe was established before AI was involved and was enforced repeatedly through to the final decision.

D2Direction & Control

Knowledge Direction & Metacognitive Control

The record shows solid workflow direction and monitoring — sensitivity analysis, scoped verification, and a sample-size catch — though the sequencing developed step-by-step rather than as a single upfront plan.

Epistemic Engagement

D3Evaluation & Reliance

Critical Evaluation & Calibrated Reliance

Critical evaluation was consistently strong: the manager discriminated among three live alternatives with distinct reasoning, rejected a weak initial model, and required a genuine counter-case before finalizing.

D4Development & Integration

Synthesis, Integration & Constructive Development

Constructive development was strong and clearly human-originated: the staged flagship-plus-conditional-store design changed the structure of the recommendation itself, not merely its wording.

D5Verification & Assurance

Epistemic Oversight & Assurance

Assurance was well-evidenced through a scoped, source-grounded verification pass covering sample size, competition coding, and causal-language calibration before the recommendation was adopted.

Responsible Ownership

D6Understanding & Accountability

Understanding, Decision Authority & Accountability

Understanding, authority, and accountability were clearly evidenced: the manager retained and exercised real authority throughout and required limitations and capital math to be checked before closing.

Collaboration

How the human and AI worked together

All three independent evaluations agreed unanimously on Co-Intelligence. The manager and AI repeatedly changed each other's thinking across the session rather than the manager simply directing and AI simply executing: the manager's reframing rebuilt AI's model, AI's gap analysis seeded the manager's staged-pilot idea, and AI's counter-case visibly shaped the manager's final reasoning.

Observed · 3 of 3The human materially shaped the AI's work

The manager's rejection of the first analytical frame caused AI to rebuild its weighting model from a repeatability-focused starting point, and the manager's staged-pilot concept caused AI to construct a new 12-month gated operating plan.

Observed · 3 of 3AI materially shaped the human's work

AI's analysis of factors the first model missed directly shaped the manager's staged-entry concept, and AI's counter-case for the non-chosen market visibly influenced how the manager framed and qualified the final recommendation.

Observed · 3 of 3The human materially developed the work

The manager originated the staged two-format, gated pilot mechanism that became the structural centerpiece of the final recommendation, not merely a constraint or approval on AI's design.

Observed · 3 of 3AI materially developed the work

AI developed the sensitivity-scenario framework, the missing-factors analysis, and the genuine counter-case — substantive analytical content, not mechanical execution of the manager's instructions.

Co-Intelligence means both the manager and AI made substantive contributions to the analysis and development, and each one materially changed the other's thinking more than once across the session. High Human Agency and Co-Intelligence are not in tension here: the manager remained intellectually consequential throughout even though AI carried a large share of the modeling, scenario development, and drafting.

Evidence

Evidence behind the result

Evidence ConfidenceModerate

Evidence Confidence describes how sufficient, direct, continuous, and well-provenanced the evidence record is. It does not change the Human Agency score. A Moderate rating here reflects the case's synthetic, single-source provenance, not a weakness in what that evidence shows.

What the record included

a pre-AI human framing notea full AI work-session transcripta verification log mapping consequential claims to sourcesa decision logthe final recommendation document

What strengthened confidence

Coverage, continuity, and trace specificity were strong: the record includes a complete chronological sequence from pre-AI framing through final close, with individual decisions traceable to specific turns and citable evidence.

What limited confidence

This is an explicitly synthetic demonstration packet, not an authenticated retrospective or prospective capture, and the verification and decision logs are derived from the same underlying work-session transcript rather than independently sourced. That caps Evidence Confidence at Moderate rather than High.

Result consistency: Strong. Three independent evaluations ranged from 84.6875–94.375, and all three placed the work in the High band.

Overall

What AgencyThread found

This market-entry process reflects High Human Agency within a Co-Intelligence collaboration pattern. The manager's contribution was strongest in framing the decision, evaluating and redirecting AI's analysis, and originating the staged entry design that the final recommendation is built around. AI contributed extensively to modeling, scenario analysis, and drafting, and repeatedly changed what the manager considered next. The most useful reading is that the manager remained intellectually consequential and in command of the decision throughout, while AI served as a substantive, reciprocal development partner rather than a passive drafting tool.

Overall: The manager remained intellectually consequential throughout the process, while AI served as a major substantive development partner.

What to notice

Two questions come first

01How much substantive human agency does the evidence support?

02What kind of human–AI collaboration produced the work?

Those are the two questions AgencyThread is designed to answer. AgencyThread does not determine whether the work is good or bad, whether AI use was appropriate, plagiarism, originality, legal authorship, intelligence, or effort. This report interprets evidence of human intellectual contribution; it does not detect AI.

Private Alpha

Imagine having this record for your own AI-assisted work.

Join the Private Pilot Talk With Us