FenoriTechnologies
Selected Builds

Institutional Memory

Preserve not only what the institution decided, but what it knew when it decided it.

We built Institutional Memory: a system that pins the full state surrounding a decision (evidence, policies, model versions, reasoning, signer) so a historical decision can be reconstructed exactly as it was made, without hindsight.

StatusReference architecture
Institutional problemHistory preserves outcomes, not what was knowable
Technical foundationDecision Objects, bitemporal state, state pinning, replay
Existing systems retainedThe decision workflows that produce the records
Human authorityReconstruction informs judgment; people decide
Evaluated onReplay fidelity, hindsight isolation
DemonstrationArchitecture walkthrough on request
DataSynthetic
DeploymentInstitution-governed AWS environment
TypeTemporal state systemDomainFinancial servicesStatusReference architectureComponents6

The institutional problem: history preserves outcomes, not knowledge.

Historical records preserve what was decided: approved, closed, escalated, filed, exited. They rarely preserve the state of the world in which the decision was made.

When a two-year-old decision is questioned, re-running the case in today’s environment answers the wrong question. Lists have changed, ownership has changed, models and policies have been replaced. What the institution needs to show is whether the decision was reasonable given what was knowable then.

Inputs and source systems.

The Decision Object is the unit of preservation. Everything the decision rested on is pinned with it.

  • Decision Objects at the moment of signature
  • The evidence, policy versions and model versions in force at that moment
  • The identity of the signer and the reasoning presented

How the system is composed.

What happens when an event enters.

Decision at T₀EvidencePoliciesSystemsModel versionsReasoningHuman signer
↓
State pinned
↓
Time passes: policies, data and intelligence change
↓
Historical question raised
↓
Replay T₀ using T₀ state
↓
Reconstruct what was known and why the decision was made

Core technical components.

  • Decision Object foundation
  • Bitemporal state
  • State pinning at decision time
  • Replay against pinned state
  • Evidence retrieval
  • Versioned policies and models

State model: two times, kept distinct

The state model is bitemporal: it distinguishes what was true from what the institution knew, and both from when it knew it. Pinning captures the decision’s knowledge state at signature; nothing that becomes known later can leak into it.

Replay reconstructs the decision from the pinned state: the evidence available then, the policies and model versions in force then, the reasoning presented then, and who signed. The institution can separately re-examine the case under today’s information, but the two questions never contaminate each other.

Example: a two-year-old clearance, questioned.

A supervisor questions the 2024 clearance of activity on Vestre Handel AS. Since then, the sanctions list, the customer’s ownership, the applicable policy and the screening model have all changed. Entities and figures in this walkthrough are synthetic.

  1. The Decision Object from March 2024 is retrieved with its pinned state: the evidence retrieved at the time, screening model v3.2, policy revision 7, the reasoning presented, and the signing officer.
  2. Replay reconstructs the case in that context. The ownership structure as known then contained no flagged parties; the screening model in force then produced no match.
  3. The institution answers the question actually asked: the clearance was reasonable given what was knowable in March 2024.
  4. Separately, the case is re-examined under today’s list and model: which do flag a party added to the ownership chain in 2025, and a new review is opened on that basis.
  5. Both answers are defensible because the two questions were never allowed to blur.

How it sits inside the institution.

Evidence and audit behavior

The pinned state is itself the evidence: an examinable, immutable account of what the decision rested on, independent of the people who made it, the systems that have since changed, and the policies that have since moved.

Where human authority remains

Replay does not re-decide. It reconstructs. Whether the historical decision stands, or new information justifies a different one now, remains an institutional judgment made by people.

Working with what already exists

Institutional Memory builds on the Decision Objects produced by the institution’s decision workflows; it does not require replacing the systems in which decisions are made.

What the system outputs.

  • A faithful reconstruction of any pinned decision in its original context
  • A clean separation between what-was-knowable-then and what-is-visible-now
  • Durable decision infrastructure that outlives staff, systems and policy changes

What it deliberately does not do.

  • It does not freeze decisions forever: a previous decision can still be reconsidered when new information appears.
  • It does not let hindsight rewrite the record: what is knowable today is kept distinct from what was knowable then.
  • It does not depend on institutional recollection: the state surrounding a decision is durable infrastructure, not memory held by people who may leave.

Hindsight should not rewrite the evidence that existed at the time.

Related builds

Built around your problem, not this one.

An institution may need this build, part of it, or something entirely different. The institution determines the outcome. The technology follows.

Discuss this outcome