Every company has two decision logs.
The first is the official one: board minutes, strategy docs, quarterly planning outputs. Formal, structured, findable.
The second is invisible: the Slack thread where someone made the call, the all-hands where the founder explained the direction change, the one-on-one where a team lead told a new hire what actually matters this quarter. These decisions are real, they shape the company, and within six months most of them are unrecoverable.
The decay rate of informal decisions
Informal decisions don't fade evenly. The people in the room remember them longest. Everyone else pieces together a version from fragments — the part of the all-hands they attended, the Slack thread they scrolled past, the secondhand summary from a colleague.
By the time a new hire joins, the unofficial decision log is already a game of telephone. By the time someone leaves, it loses another node. By the time a decision needs to be revisited, the original rationale is gone — and the team either relitigates from scratch or makes a call that silently contradicts what was already decided.
This isn't negligence. It's what happens when the cost of capturing informal decisions is high relative to the benefit at the moment the decision is made.
What AI can and can't do here
AI won't automatically capture your informal decisions. That's not how it works. The signal still has to be somewhere.
But the threshold for "somewhere" is lower than most teams think. A Slack channel where decisions get briefly summarized. An L10 notes doc where the IDS resolution gets a sentence. A voice memo that gets transcribed. None of these are formal decision logs — but they're enough for an AI with good retrieval to reconstruct what was decided and why.
Freddy doesn't require your team to change how they make decisions. It requires that decisions leave enough of a trace to be found.
The practical shift
The teams that retain institutional memory well don't have better documentation cultures. They have lower friction for capturing the things that matter.
When someone can ask "what did we decide about the partner discount structure?" and get an accurate answer from a Slack thread that was summarized six weeks ago, they stop needing everyone who was in the room. The decision becomes part of the company's operating context — retrievable by anyone, not just the people who were there.
That's the shift. Not documentation for its own sake. Context that stays accessible as the company grows and the room changes.
BrainGem builds Freddy, an AI that lives in Slack and learns your company's operating context — so decisions that don't get formally written down don't get permanently lost.
Top comments (0)