Chat before Canvas: a workflow for designing with AI
Design is hand-to-mind. You make something, look at it, feel something is off, move it, look again. The solution emerges from the doing. You don't know what you're looking for until you see something that isn't there yet.
AI is mind-to-hand. It needs you to know what you want before you ask. The thinking has to happen before the making, not through it.
Not a criticism of AI just a different cognitive sequence. And for certain design tasks, it just feels backwards.
When you skip the thinking and go straight to output, you don't get faster work. You get a tax. Every decision lands at once. You're not building understanding through the work, you're auditing someone else's interpretation of your intention. That's neither fast nor efficient.
This workshop is about not paying that audit tax.
What you'll do
Two passes on the same loose idea. The first is what most of us do, straight into the AI, generate output, try to figure out what it decided. The second is the chat-first workflow, sharpen the idea through conversation, generate from a brief you actually wrote, then interrogate what comes back.
The point isn't to judge which output looks better. It's to feel the difference between auditing AI's decisions and reading your own work back.
The session runs 60 minutes plus an optional 10 minutes of open Q&A afterwards. Stay on if you want to keep the conversation going. Leave on time if 60 is what you have.
What you'll leave with
A workflow you can use Monday morning. A Google Doc with your sharpened thinking and refined output. A recording of the session. A Q&A transcript by email. And a clearer sense of where you're being a designer in your AI workflow, and where you're being an detective, digging for clues.
Who this is for
Designers using AI tools and feeling the cognitive friction. Open across seniority. Adjacent roles welcome, PMs prototyping their own ideas will recognise the same shift.
Honest framing
I'm working through this in public, like everyone else is. What I have is a workflow that's been holding up and a framework for why. We'll walk through it together. Some of what I teach today I'll teach differently in three months because the people in this room will have shown me where it breaks.
Bring
A real loose idea or design problem you're working on. Doesn't need to be fully scoped. Looser is fine.