← Back to the proposal

the thinking · design

The Design Problem

The problem with AI-generated copy isn't that the model writes badly. It's that it writes generically.

The problem with AI-generated copy isn’t that the model writes badly. It’s that it writes generically. It knows Cub Scouting in general and nothing about your pack. And families notice: over half of consumers report feeling less engaged with content they suspect is AI-generated[^1].

The answer is the pack’s own knowledge: the things only the pack knows about its families and what each new family tends to need. That knowledge lives in volunteers’ heads, almost never written down. When a Cubmaster steps down, most of it leaves with them.

Any AI already knows this

  • What earning the Bobcat rank requires: the Scout Oath, the Law, the sign, the salute.
  • That Pinewood Derby is a Cub Scout tradition with rules about car dimensions and weight.
  • What the Scout Oath and Law say, word for word.
  • That Blue & Gold is a February banquet celebrating Scouting's birthday.

Only your pack knows this

  • That the Patterson family is shy in groups.
  • That the Cub Scout handshake gets demonstrated at the September pack meeting.
  • That Dan Boyle takes the building side of Pinewood Derby far too seriously.
  • That the Hendricks family has been here three years and doesn't need the basics anymore.
The left column is what the model arrives with. The right is what Pack Memory holds.

So we need one place where that knowledge can be captured and edited, in plain language, by the people who already hold it. We call that place Pack Memory: the pack’s institutional memory, finally written down.

See what it produces →

[^1]: How consumers interact with AI vs human-made content. Bynder press release / research study, April 2024. N = 2,000 UK and US general consumer panel; respondents shown matched articles by a copywriter and ChatGPT. 52% reported feeling less engaged with content they suspected or knew was AI-generated. Cited here as directional evidence of consumer AI-detection sensitivity, not as a precise estimate for parent-facing pack communications. Bynder