Softlander Book a call
Agent evaluation and assurance

Agent regression testing

Detect changes that break previously accepted agent behaviour

A model, prompt, memory policy or connector has changed. The decision is how to detect changes that break previously accepted agent behaviour. 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 release managers and evaluation owners. A useful outcome is an explicit account of what was examined, what remains uncertain, and which decision the evidence can support.

Book a 30 minute readiness callBrowse this capability group

The release evidence relationship

  1. Starting evidenceVersioned baselines

    Establish the basis for protected constraint.

  2. Decision boundaryBaseline revision

    Separate a changed requirement from a failed case.

  3. Decision artefactRegression suite

    Retain the scoped finding and its unresolved dependencies.

Proposed working model. The customer retains the release decision.

Decision

Frame the decision around protected constraint

Questions and the evidence they need

Criterion / evidenceProtected constraint
Inspect versioned baselines for the relevant constraint or observation.
Criterion / evidenceBaseline revision
Inspect accepted case results for the relevant constraint or observation.
Criterion / evidenceAction comparison
Inspect change notes for the relevant constraint or observation.
Read each criterion against the available evidence before deciding that additional work is needed.

The decision is how to detect changes that break previously accepted agent behaviour. This is relevant to release managers and evaluation owners when a model, prompt, memory policy or connector has changed. 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 an improvement in answer quality introduces a new unauthorised write or removes a necessary escalation. 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

Protect required behaviour across each revision

Relationships within release evidence

Source or conditionProtected constraint
Retain approval and correct-target requirements.
Crossing or choiceBaseline revision
Separate a changed requirement from a failed case.
Consequence or decisionAction comparison
Compare tool effects as well as generated responses.
Keep case coverage linked to the release question. Each relationship needs evidence in the actual scope.

What this specialty examines

A protected behaviour is a requirement that should survive a change, not necessarily an exact response string. Preserve constraints such as requiring approval, selecting the correct account and avoiding duplicate writes. Allow legitimate variation in phrasing or reasoning that does not alter the requirement.

The review should replay protected behaviours against both revisions and inspect changes in tools, state and final responses. 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, stable regression cases protect known behaviour but need refresh when workflows legitimately change. 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 protected constraint

Illustrative working format

BasisVersioned baselines
Keep workload, quality, human effort, and uncertain inputs visible in comparisons.
ContextAccepted case results
Retain the observed conditions and distinguish gaps from contrary evidence.
ObservationChange notes
Name the custodian and the conditions requiring a refreshed version.
DependencyRelease configurations
Match this description to the deployed or proposed boundary being examined.
Illustrative preparation record. Availability, revision, and handling need confirmation.

Start with versioned baselines, accepted case results, change notes and release configurations. 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 an improvement in answer quality introduces a new unauthorised write or removes a necessary escalation 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 protected constraint through to an accountable finding

From the scoped material to a decision record

  1. Establish conditionsVersioned baselines

    Separate hard constraints from graded preferences.

  2. Examine the relationshipBaseline revision

    Apply the scoped check to action comparison.

  3. Retain the findingRegression suite

    Record the evidence, open dependencies, and the owner’s next decision.

Proposed sequence; resolve interpretation or access gaps before relying on the next step.

Prepare the evaluation cases

Work with release managers and evaluation owners to assemble versioned baselines, accepted case results, change notes and release configurations. 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

When a case changes, record whether the product requirement changed or the old expectation was mistaken. Do not overwrite a failed baseline simply to make the new release green. Keep the rationale and approval for new expectations separate from the observed comparison.

Choose the release or evaluation change

Use the reviewed cases to produce a regression suite and change report separating fixes, losses and unresolved differences. 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 regression suite usable by its owner

Regression suite

Illustrative working format

Annotated fieldRegression suite
Identify the assessed workflow, revision, and decision boundary.
Annotated fieldFixes
Link supporting and contrary observations to inspectable evidence.
Annotated fieldLosses
Name the unresolved condition, next action, and accountable owner.
Illustrative artefact structure, not a completed engagement or an observed result.

A useful decision artefact is a regression suite and change report separating fixes, losses and unresolved differences. 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 action comparison through a constructed exception

A constructed case and its decision paths

Illustrative scenario

  1. SituationProtected constraint

    A shorter prompt omits confirmation before a subscription change.

  2. Question to examineBaseline revision

    Inspect whether confirmation vanished from text or actual control flow.

  3. Conditional next actionAction comparison

    Retain the regression if required authorisation was removed.

Fictional example. No executed test, customer engagement, or measured outcome is implied.

Consider a fictional situation: a shorter prompt reduces verbosity but causes an assistant to skip confirmation before changing a subscription. The scenario is illustrative, with no client engagement, executed test or measured improvement implied.

For the shorter prompt, inspect whether the confirmation step disappeared only from text or from the action sequence itself. If authority checks still hold elsewhere, the consequence differs. The regression report should connect the changed behaviour to the actual control boundary.

Scope and next action

Keep the conclusion within selected task cases

Confirm fit before the next commitment

Included boundaryProtected constraint
This review is limited to selected task cases.
PrerequisiteVersioned baselines
The relevant owner must confirm access and the intended operating conditions.
Claim limitRegression suite
The customer retains the release decision.
Bring the triggering change and existing evidence; resolve the dependency that could alter the 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 a model, prompt, memory policy or connector has changed; 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.

Read about this offer

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.

Read about this offer

AI Long-Horizon Agent Evaluation

Examine what changes when an agent keeps working. Evaluate accumulated state, recovery and goal adherence across extended tasks.

Read about this offer

Continue exploring

Related insights and resources

Whitepaper16 MIN READ

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.

Whitepaper15 MIN READ

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.

Whitepaper5 MIN READ

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.

A little more clarity

Questions to consider.

Start with the questions that shape a better decision.

When should an expected result be updated?

Update it when the intended requirement changes or review establishes that the original expectation was wrong. Preserve the prior result and the reason for revision. A failing release alone is not grounds to rewrite the expectation.

Which baseline records should survive a release change?

Start with versioned baselines, accepted case results, change notes and release configurations. Record their revision, owner and relationship to the intended workflow. Missing context should be identified before interpreting outcomes; a sanitised fixture is useful only when its relevant differences from the operating case are understood.

How should changed tool behaviour be compared?

Replay protected behaviours against both revisions and inspect changes in tools, state and final responses. Keep task correctness, permitted action and final state distinct. Review scoring disagreements against the agreed rubric. Include valid alternatives so the cases do not reward superficial sequence matching or rejection of every difficult request.

What should a regression report preserve?

The proposed output is a regression suite and change report separating fixes, losses and unresolved differences. Include the selected conditions, supporting observations, unresolved differences and next decision. A reviewer should be able to follow each important conclusion back to its evidence and understand the limits without depending on an undocumented explanation from the author.

When does a product change invalidate the baseline?

A changed product requirement can invalidate the corresponding expected outcome, but not automatically every baseline. Record the authorised change, preserve historical results and rerun the behaviours that should remain protected.

What if existing evidence already answers the question?

Review its scope, provenance and configuration before proposing new work. Existing work from release managers and evaluation owners may already be sufficient. In that case, capture the decision, unresolved limitations and the change that would require another review. Additional assessment should address a real gap.

What would make the findings inconclusive?

Missing versioned baselines, accepted case results, change notes and release configurations can prevent a reliable interpretation, depending on the case. Record the specific missing observation and its implication. The owner can supply evidence, narrow the claim or defer the dependent decision; an incomplete case should not silently be counted as a pass.

Start with the decision you need to make.

Tell us about the agent workflow you want reviewed, and where you need greater clarity. Scope and availability are confirmed together.

Book a 30 minute readiness call Get the agent release checklist

Connected disciplines

Golden trace development

Define reference traces without prescribing one brittle reasoning path. The review connects expert demonstrations, tool contracts, permitted alternatives and terminal-state evidence to a bounded review of evaluation evidence, giving the release owner a clearer basis for the next action.

Adversarial agent evaluation

Understand how an agent responds to deliberately difficult task conditions. The review connects authorised attack scenarios, protected actions, synthetic records and escalation rules to a bounded review of evaluation evidence, giving the release owner a clearer basis for the next action.

Tool-use evaluation

Verify that tool actions match the request and the agent's authority. The review connects tool schemas, permission rules, request examples and tool-side execution logs to a bounded review of evaluation evidence, giving the release owner a clearer basis for the next action.