Human–agent interaction evaluation
Design evidence for the moments people approve, correct or interrupt an agent
Human review is listed as a control without testing the reviewer's actual experience. The decision is how to design evidence for the moments people approve, correct or interrupt an agent. The work begins with your actual workflow and looks for the evidence that separates intended behaviour from a consequential exception. It is particularly relevant to product designers and operations leaders. A useful outcome is an explicit account of what was examined, what remains uncertain, and which decision the evidence can support.
The release evidence relationship
- Interaction flows
Establish the basis for review context.
- Reviewer choice
Support refusal and correction before execution.
- Interaction evaluation map
Retain the scoped finding and its unresolved dependencies.
Decision
Frame the decision around review context
Questions and the evidence they need
- Review context
- Inspect interaction flows for the relevant constraint or observation.
- Reviewer choice
- Inspect approval screens for the relevant constraint or observation.
- Interruption
- Inspect reviewer instructions for the relevant constraint or observation.
The decision is how to design evidence for the moments people approve, correct or interrupt an agent. This is relevant to product designers and operations leaders when human review is listed as a control without testing the reviewer’s actual experience. Name the affected workflow, the consequence of a wrong decision, and the owner who can act on the result. The initial discussion should establish which evidence is already available and which uncertainty is material enough to investigate.
The specific failure hypothesis is that a reviewer approves an action without seeing changed parameters or relevant uncertainty. It remains a hypothesis until examined in the deployment. Existing internal tests or operating records may already resolve it; identify the uncovered question before commissioning additional work.
Mechanism
Approval is meaningful only with the action context
Relationships within release evidence
- Review context
- Expose the exact changed fields and uncertainty.
- Reviewer choice
- Support refusal and correction before execution.
- Interruption
- Verify timeout and cancellation behaviour.
What this specialty examines
Human interaction evaluation should examine what the reviewer knows at the moment of intervention. Show the proposed action, changed parameters, supporting evidence and meaningful uncertainty. A reviewer cannot make an informed decision from a reassuring summary that omits the actual consequence.
The review should walk through approval, refusal, correction, timeout and interruption cases with representative roles. Identify the normal route, relevant exceptions and the authoritative record of the outcome. Inspect the relevant business-system record, enforcement event or process evidence alongside any reported status. Keep incomplete cases visible, with an explanation of what their absence prevents the owner from concluding.
The tradeoff to make explicit
For this decision, extra confirmation can strengthen control while creating review fatigue and abandonment. Compare the available design or operating choices against the same intended outcome. If one option removes a required control or changes the task boundary, describe that difference before treating its apparent simplicity, speed or lower effort as an improvement. The decision should expose the tradeoff instead of burying it in an aggregate score.
Preparation
Connect the working inputs to the boundary being examined
Evidence packet for review context
Illustrative working format
- Interaction flows
- Identify provenance and the specific decision this material can inform.
- Approval screens
- Connect the accepted action parameters to grant, amendment, expiry, and withdrawal.
- Reviewer instructions
- Identify the accountable role and verify its decision rights or availability.
- Intervention records
- Establish permitted access, provenance, and which task conditions it represents.
Start with interaction flows, approval screens, reviewer instructions and intervention records. Identify the owner, revision and operating context of each source. Sanitised examples may support initial design, but differences from actual conditions should remain explicit. Where direct access is unsuitable, a customer-controlled evidence or execution route may be considered, with agreed handling and retention.
The scope also needs the release owner, agreed expectations for a representative task case, and the authority to inspect the relevant boundary. Specify what evidence would distinguish the failure in which a reviewer approves an action without seeing changed parameters or relevant uncertainty from acceptable behaviour. If stakeholders disagree about intended behaviour, resolve the interpretation before classifying the observed difference as a defect.
How the work proceeds
Carry review context through to an accountable finding
From the scoped material to a decision record
- Interaction flows
Separate hard constraints from graded preferences.
- Reviewer choice
Apply the scoped check to interruption.
- Interaction evaluation map
Record the evidence, open dependencies, and the owner’s next decision.
Prepare the evaluation cases
Work with product designers and operations leaders to assemble interaction flows, approval screens, reviewer instructions and intervention records. Select a normal task, a consequential exception and a case with insufficient information. Identify the relevant requirement and expected result for each, then mark which outcomes need human adjudication. The preparation output is a case packet that separates blocking behaviour from graded quality preferences before any comparison is run.
Run and review the comparisons
Test refusal and correction as carefully as approval. Confirm that editing an action invalidates stale approval, that timeouts do not silently authorise execution, and that a reviewer can stop or escalate work. Record usability obstacles that encourage habitual approval without inspection.
Choose the release or evaluation change
Use the reviewed cases to produce an interaction evaluation map and proposed usability and control acceptance criteria. Distinguish a product defect from a missing test, an ambiguous requirement or an unreliable measuring method. For each, name the proposed correction and the case that would check it. The release owner can then decide whether to change the implementation, improve coverage or restrict the intended use while a specific gap remains.
Indicative artefact
Make the interaction evaluation map usable by its owner
Interaction evaluation map
Illustrative working format
- Interaction evaluation map
- Name the missing evidence, responsible owner, and condition for revisiting the decision.
- Usability
- Link supporting and contrary observations to inspectable evidence.
- Control acceptance criteria
- Define the required observation and who resolves disagreement about acceptance.
A useful decision artefact is an interaction evaluation map and proposed usability and control acceptance criteria. Supporting material may include selected cases, configuration or process records, evidence references, a disagreement log and prioritised actions. Retain the case selection rationale and material disagreements alongside the result; otherwise the result may be mistaken for complete coverage.
Illustrative scenario
Examine interruption through a constructed exception
A constructed case and its decision paths
Illustrative scenario
- Review context
A supervisor approves a summary that hides an additional changed field.
- Reviewer choice
Expose the complete payload beside the approval decision.
- Interruption
Invalidate approval when the operation includes an unreviewed field.
Consider a fictional situation: a supervisor sees a summary saying account update approved while the underlying change includes an additional field. The scenario is illustrative, with no client engagement, executed test or measured improvement implied.
In the account-change example, compare the approval screen with the final parameters. If an additional field appears after approval, the system should require a new decision or prevent the change. The design question is how to preserve informed control without unnecessary repeated prompts.
Scope and next action
Keep the conclusion within selected task cases
Confirm fit before the next commitment
- Review context
- This review is limited to selected task cases.
- Interaction flows
- The relevant owner must confirm access and the intended operating conditions.
- Interaction evaluation map
- The customer retains the release decision.
A result applies to the recorded configuration and selected cases. It does not establish universal agent competence or authorise release. Any engagement depends on an agreed scope, qualified team, authorised access and confirmed availability. Bring the affected workflow, its current configuration or process, the triggering change and the evidence already available. The immediate scope question is prompted by the situation in which human review is listed as a control without testing the reviewer’s actual experience; confirm which missing observation would change the owner’s next action.
Ways to engage
Related engagement scopes
AI Agent Evaluation Evidence Diagnostic
Find the gaps between your release decision and the evidence behind it. Start with one agent workflow and a bounded set of questions.
AI Agent Evaluation Foundation
Turn scattered tests into a repeatable release discipline. Connect the behaviours that matter to datasets, evaluation criteria and evidence your team can maintain.
AI Long-Horizon Agent Evaluation
Examine what changes when an agent keeps working. Evaluate accumulated state, recovery and goal adherence across extended tasks.
Continue exploring
Related insights and resources
Engineer the harness, not just the prompt
Define the contracts around an agent's decisions: what state it trusts, which actions can execute, how interrupted work resumes, and what evidence an operator needs to recover safely.
What would justify releasing this agent?
Replace a single pass rate with a reviewable chain of requirements, cases, observations and judgements. Distinguish task quality, control failures and uncertainty before increasing an agent's exposure.
The evidence behind the go-live decision
A working paper for leaders who need to explain why an agent workflow is ready for its next release. Includes a proposed evidence-pack structure, decision table and review agenda.