The Problem
Decisions live everywhere, and nowhere
Today, corporate decisions live nowhere in particular. They appear in fragments, scattered across a multitude of tools that were never designed to hold them. A discussion begins in Slack, continues via email, is clarified orally during a meeting, is settled over the phone, and then sometimes leaves an indirect trace in a Jira ticket or a ServiceNow form. The action itself is executed elsewhere still - in a terminal, a cloud console, or an automated system - while applications log what they can in files designed for diagnostics, not for proof. At the moment, this dispersion is not a problem. Teams understand each other, the decision flows, the organization moves forward.
The question always comes later
The difficulty only appears later, when one tries to establish what actually happened. At that point, there is no single point to turn to. One must reconstruct. We gather what remains: messages extracted from Slack, screenshots attached to emails, PDFs exported from business tools, tickets modified multiple times, technical logs that are difficult to interpret outside their original context. Very quickly, it becomes clear that this material is incomplete. Some discussions were never written down, some exchanges took place orally, some messages were deleted, some logs were purged, some people are no longer there. What is missing can no longer be recovered.
Reconstruction replaces proof
Faced with these absences, reconstruction inevitably becomes interpretive. Gaps are filled with memories, assumed intentions, and reasoning formulated after the fact. A coherent story takes shape, often in good faith, sometimes under pressure, but always influenced by the context in which the question is asked. This narrative may seem solid, but it no longer constitutes proof. The elements produced - screenshots, PDF exports, isolated emails - do not carry their own integrity. They do not allow an independent third party to verify that they have not been modified, selected, or taken out of context. They ask to be believed, and as soon as a piece of evidence asks to be believed, it ceases to be binding.
Time works against certainty
Over time, this fragility only worsens. Tools evolve, formats change, systems are updated, and retention policies erase what was not intended to last. Memories alter, certainties shift. Six months later, or sometimes a year later, the decision only exists as a narrative the organization is capable of producing about itself. This narrative may be honest, precise, and detailed, but it remains vulnerable to doubt because it relies on a late reconstruction of a reality that was never captured.
Tools were never designed for proof
The problem is not that companies lack tools. On the contrary, they use too many, each optimized for a specific function. Slack facilitates discussion, email structures communication, Jira organizes work, ServiceNow frames processes, technical systems execute, logs observe. But none of these tools is intended to freeze the moment a decision becomes irreversible, nor to attest that an authority declared something at a specific time. The decision traverses systems without ever being fully inscribed in them. It exists in action, but not as an autonomous, independently verifiable fact.
Explaining is not proving
When the question is finally asked - often long after the events - the organization can no longer show what happened. It can only explain it. And explaining, as rigorously as possible, is never equivalent to proving. This is not a problem of discipline, nor of method, nor of goodwill. It is a structural problem. As long as it remains invisible, organizations will continue to believe they can explain later, only to discover, too late, that explaining is not proving.