For brand teams that want answers with their taste in them.

We run the room where your team puts its taste into words. Then we keep it current for your AI.

Workshops we run with your brand team, in a room, with cards on the table. Three formats, from 90 minutes to two days, and a monthly review.

What you see today

Your AI gets most of the brand work right. The last part is taste nobody wrote down.

The facts hold. It misses how you pitch, decide and say no.

Everyone has a seat. It answers as a stranger, and the checking never stops.

Someone else's voice →

What a brand book cannot hold

A brand book or a voice prompt holds what you already wrote. Taste is unwritten knowledge: the judgment behind the work, which nobody wrote down.

Unwritten knowledge →

Fast layers make your AI current. Slow layers make it sound like you.

How context decays →

In the room, your team puts the reasons into words, with cards on the table. You bring a real problem, and the room writes down one thing it always knew and never wrote.

What you get

  • Answers that sound like you, with the right numbers.
  • What your AI needs to know about you, in your own words.
  • One corner lit: a piece of unwritten knowledge, written down and ready to use.
  • Context files your AI can use from day one.
  • A memory that stays current as people and priorities change.

Where to start

If your AI answers as a stranger, start with the session. It is best when you want to start the conversation in a department.

See the session →

If one or two people get answers that sound like the brand and the rest do not, start with the mapping.

See the mapping → Where are you? →

Questions brand teams ask

Can we not just give the AI our brand guidelines?

Give it the guidelines. They help with what is written. A custom GPT with your brand guidelines attached knows your fonts, not your judgment.

Who should be in the room?

The people who make the work and judge it, and the brand owner who can act on what the room finds.

What happens to our material?

Not everything belongs in the record. You decide what goes in and what stays yours. A model cannot leak something it cannot see.

Put your taste into the answers.

Tell us where your AI sounds like someone else. A short call. Then we both decide if there is a fit.