When the Hallway Stops Talking to the Machine

Important conversations should be turned into durable artifacts before the memory evaporates.

The Spark

Yesterday, two unrelated conversations got me thinking.

The first was practical. My manager had just made a decision and asked if I wanted him to send me an email “for the record.” In this case, the decision needed to be documented in an Azure DevOps work item, so I told him not to bother. The decision would already be remembered.

A little later, a colleague made an observation about return-to-office initiatives and Microsoft Copilot. As more conversations move back into conference rooms, hallways, lunch tables, and impromptu whiteboard sessions, Copilot is increasingly absent from some of the most important discussions. It cannot summarize what it never saw.

Together, they raised an interesting question: how does the knowledge we capture change based on where we work?

The Accidental Golden Age of Organizational Memory

For a brief period, many organizations became better at creating institutional memory. Remote work pushed conversations into digital spaces:

• Teams chats
• Emails
• Shared documents
• Recorded meetings
• Ticket systems
• Collaborative workspaces

What previously vanished into the air now left a trail.

Project decisions that once might have happened in conference rooms became Teams threads. A design debate became comments on a document. Questions and answers accumulated in searchable systems instead of living exclusively inside someone’s head.

During COVID, we thought of these tools as communication platforms that allowed business to continue. Today, AI systems increasingly use them as memory.

What Happens When Conversations Move Back Offline?

As organizations return to the office, some amount of communication naturally shifts back to face-to-face interactions.

This is not necessarily bad.

Humans have been sharing knowledge through conversation for thousands of years. Many important forms of collaboration still work better when people are physically present. Mentoring, trust building, brainstorming, and nuanced problem solving often benefit from being in the same room.

The challenge is that these conversations are frequently ephemeral.

• A decision made during a hallway conversation may never reach a document.
• A design assumption discussed over lunch may never reach a ticket.
• A project rationale sketched on a whiteboard may disappear when the board is erased.

The people involved may remember. The organization may not.

The Question Is Bigger Than AI

Before Copilot, organizations struggled with the same issues that AI now exposes:

• Why was this decision made?
• What alternatives were considered?
• Who approved the change?
• What assumptions existed at the time?

When knowledge is captured, modern systems can often retrieve it. When knowledge is never recorded, even the most advanced AI has nothing to work with.

A Working Hypothesis

This leads to a hypothesis worth exploring:

Organizations that successfully adopt AI will not necessarily be the ones with the most sophisticated models. They will be the ones that consistently convert important conversations into durable artifacts before the memory evaporates.

The artifact does not have to be an email. It might be:

• a DevOps work item
• a design document
• meeting notes
• a decision log
• a wiki page
• an architecture record

The format matters less than the existence of a shared, discoverable record.

This does not mean every conversation needs to become a transcript. That would be exhausting, and probably counterproductive. The practical habit is smaller: when a conversation changes a decision, an assumption, a commitment, or a plan, leave a trace where the next person and the next system can find it.

The New Cost of Forgetting

For years, undocumented decisions created friction for future employees trying to understand the past. Today, every undocumented decision is invisible to the organization’s AI systems.

As return-to-office strategies continue to evolve, the most successful organizations may not be the ones that choose remote work or office work. They may be the organizations that recognize an old truth: Important conversations are only valuable for as long as someone remembers them.

The moment memory becomes a business dependency, documentation becomes a strategic capability.

AI did not change that reality. It made it impossible not to see. And I believe AI will be one of the tools that makes this manageable.