FG—01AI Design Field Guide

Search the field guide

Search chapters, principles, patterns, worksheets, and glossary terms

The argument

Principles

Ten commitments, grouped by when they bind: before the interaction, during it, when the AI is wrong, and over time. Every pattern in this guide exists to serve at least one of them.

Before the interaction

  1. PR—01Set the contractTell people what the AI can do — and how well — before they invest.
  2. PR—02Earn the first promptAn empty text box is a test most people fail — seed it, situate it, or lose them.

During the interaction

  1. PR—03Show the workExpose the process, not just the answer — legibility is where trust actually comes from.
  2. PR—04Keep the human seniorThe AI drafts, proposes, and executes — the human directs, approves, and can always grab the wheel.
  3. PR—05Match effort to stakesSpend interface, latency, and ceremony in proportion to what the task is worth.

When the AI is wrong

  1. PR—06Fail in the openWhen the system is unsure or wrong, say so plainly — a visible failure is recoverable, a hidden one is a betrayal.
  2. PR—07Make recovery cheapDesign for the wrong answer: undo, retry, refine, and branch should cost seconds, not sessions.

Over time

  1. PR—08Make memory legibleIf the system remembers, people must be able to see, edit, and delete what it holds.
  2. PR—09Feedback must matterNever ask for feedback you won’t visibly act on — close the loop or drop the button.
  3. PR—10Never surprise with changeThe system may improve, but it may not shape-shift — announce change, preserve continuity, degrade predictably.