By problem

A memory that stays current as people and priorities change.

It cheerfully suggests the thing you already tried in 2023, because nothing told it otherwise.

Written down does not mean up to date. We keep the context current for your AI, with you.

You have seen this

You ask for ideas. The first idea on the list is one your team already tried and dropped.

It is working off a deck three years old, not from where you and your strategy actually are today.

The first pilot worked. Six months on, the files it reads are stale and nobody owns them.

The most expensive lesson is the one you pay for twice.

Why it happens

Where does the lesson from that attempt sit today?

Where your company memory sits

  1. Unsaid in a head why you priced it that way
  2. Verbalized said once, in a meeting what you decided over lunch
  3. Documented written, out of your AI's reach last year's offer, in a folder
  4. Connected your AI reads it and acts on it examples & guidelines AI can use

Your AI can only see the last step.

dark context noun

What your organisation knows but never wrote down. The biggest part, and the part a general AI model cannot see.

What is dark context →

What has already been tried and abandoned is part of your institutional memory.

The model doesn't know that it doesn't remember. An assistant answers anyway.

Institutional memory →

What was written can go gray: still findable, no longer true.

Dark context keeps you honest: you know you are missing something, so you ask, you check, you leave room for doubt. Gray context does the opposite.

Gray context →

Hand an organization's knowledge to an AI and part of it is stale by next week, part will still be true in ten years.

Context decay →

What to do

Each of the three formats is a workshop that we run with your team, in a room, with cards on the table. Each one takes your memory one step up.

Find the row that sounds like your team today.

StageYour AI's answersNext step
Low lightgenericThe session
Corners liton pointThe build, then the monthly review
Kept lityoursKeep it current
See all five rows →

Questions

Can we not do this ourselves, in our wiki?

You can have a perfectly documented, easy-to-find playbook that is completely gray, because it describes how the company worked two years ago. It passes every check, and it is still wrong. The page records the decision. The person carried the why. In the room, people say the why out loud and write it down.

Can we stop after one format?

Yes. Each of the three formats ends with a named output that you keep. The review is there for as long as people and priorities keep changing.

Tell us what you already tried.

Book a call, and bring one answer your AI got wrong because it did not know your history.