Home / Blog / Decision reconstruction

Decision evidence · 2026-08-09

Why replaying yesterday's AI decision with today's memory creates a false explanation.

The replay can look perfect. The answer can even be better. That is exactly the problem: it explains what the agent would say now, not what the agent could have known then.

The failure mode

Current truth is not historical truth.

Suppose an underwriting agent reviewed a company on March 1. The source system said annual revenue was $12 million. On March 8, finance corrected the record to $8 million and backdated the corrected value to January 1.

If an investigator reruns the March 1 prompt against the database today, retrieval returns $8 million. The reconstructed answer may cite the corrected record, use the same model, and follow the same instructions. It still cannot explain the original decision, because the correction did not enter the system until March 8.

# What the fact says about the world
event_time:     2026-01-01

# When the system could first know this version
ingestion_time: 2026-03-08

# A March 1 reconstruction must exclude it
reconstruct(as_of="2026-03-01")  → revenue = $12M

Why replay is insufficient

A prompt, trace, and current database are three different records.

A prompt records instructions

It does not prove which source versions were available, admitted into memory, or excluded by policy at the moment of action.

A trace records execution

It can show calls and timings, but a reference to a mutable document or row may resolve to different content later.

A current database records what is believed now

It usually optimizes for present truth. Corrections, deletions, re-indexing, and permission changes can overwrite the boundary the original agent saw.

A reconstruction records the historical aperture

It asks which exact facts, source versions, policies, identities, tools, and approvals were knowable and permitted at the decision time.

The minimum defensible record

Freeze the boundary, not just the output.

A useful decision receipt should bind the result to enough evidence to answer six questions later:

What did the agent receive?

The prompt or instruction version, model identity, parameters, recalled facts, source versions, and relevant tool results.

When was each fact knowable?

Both event time and ingestion time, plus the validity interval and any supersession relationship.

Who and what were authorized?

Workload identity, tenant and barrier scope, policy version, approval state, and exact downstream request.

What happened next?

The action attempted, permit issued, side effect observed, reviewer intervention, and linked outcome.

What is missing?

Declared capture gaps. A partial record should say it is partial instead of presenting a neat replay as complete evidence.

Can the record be verified?

Content addresses, signatures or trusted digests, append-only sequencing, and exportable material for independent review.

A practical test

Correct one fact and measure the blast radius.

Pick a synthetic or approved workflow and record a consequential decision. Then introduce a late correction to one source fact. A governed evidence layer should be able to:

Keep present recall current

New work should use the corrected fact and exclude the superseded value.

Keep historical recall honest

A reconstruction before the correction arrived should still surface the older fact that was actually knowable then.

Find exposed decisions

The correction should reveal which decisions directly or transitively depended on the superseded source.

Preserve remediation evidence

Owners, review status, corrected outcomes, and closure should remain connected to the original record.

That test is more revealing than asking whether a model can reproduce similar prose. It tests whether the system can distinguish present truth from the information boundary that shaped an earlier action.

Claim boundary

Reconstruction is evidence, not certainty.

A reconstructed record cannot prove context the system never captured, guarantee deterministic model behavior, establish legal causation, or make a workflow compliant. It can make the boundary inspectable: what was recorded, what was knowable, what was permitted, what changed, and where the evidence is incomplete.

See the lookahead-bias mechanism → · Read the governed memory guide → · Inspect the source ↗

Prove what the agent knew when it acted.

Scope one workflow →