← All classes
CLASS 1May 26 · Course Introduction

"We truly believedesigners lead, and AI follows."

Dominik Heinrich
Taught byDominik HeinrichDesign Intelligence & Tech Experiences, Coca-Cola; Co-founder, Creative AI Academy
With Tony Jones, ex-ECD McCann; Co-founder, Creative AI Academy

The class in brief

Orientation night: Dom Heinrich and Tony Jones laid out the course's central bet before touching a single tool. AI is not intelligent, it is a predictive algorithm, and the semester exists to build the design judgment that keeps its output from disappearing into the crowd. They named the three pillars, walked the required tool stack, and introduced the twelve guest lecturers who carry the rest of the semester. After this page you can state the course's three pillars from memory and name the two or three tools that actually cover your own repeated workflow.

The night at a glance

Why this matters · 6:00 PM

These machines are not intelligent. They're just predictive algorithms, mathematics.

Before the syllabus, the thesis: technology is not a disruptor, it is an accelerator, the same relationship humans have had with every tool since the stone age. Dom's flattest line of the night is the one the whole semester rests on: the model is only as good as the person who uses it. Class started with a shared Miro board the students had been filling in since the afternoon: day-to-day tasks, tools already in use, the biggest challenge, one goal for the workshop.

3
pillars carrying the whole semester: Prompt Design, Thinking Design, Human-AI Relationship.
80%
attendance required for the certificate. Watching a recording does not count, and recordings live 30 days.

The framework · 6:30 PM

The three pillars run as one loop

01

Prompt Designinteraction techniques

Workflows, tool limits and capabilities. The first half of the semester lives here: the technologies, the language, the habits of asking well.

02

Thinking Designthe method, not the tool

Innovative processes, infinite design methodology. Dom's own framework, taught in full later in the course, moves the work from design thinking to thinking design.

03

Human-AI Relationshipthe final project lives here

Collaborative systems, autonomous agents, the capstone group project. The semester's last stretch and its real subject.

The method underneath all three: humans think, machines generate, in an infinite loop. Not one fixed output, an ever-evolving, adaptive design. Dom will teach a full class on it later in the semester, but the loop starts governing the room tonight.

The craft · 7:15 PM

Go deep on two or three tools. Skip shallow on a hundred.

The required stack for the semester: ChatGPT Plus, Krea Pro (free with a course code), and Adobe Firefly through Pratt's Adobe Creative Cloud login. Worth exploring beyond that: Gemini, Claude, Suno, HeyGen, Ideogram, and Geoff Gibbins's Corrix browser plugin, arriving with him in Class 3. Thirteen guest lecturers were named for the semester, each teaching a different slice of the craft, from strategic foresight to synthetic personas to node-based Krea workflows.

Public prompt libraries are a dead end for the same reason a hundred shallow tools are: if everyone uses the same prompt, everyone gets the same output. A couple of tools that fit your specific workflow steps beat a Swiss-army kit that fits none of them precisely.

12
guest lecturers carried the rest of the semester, from strategic foresight to bias illumination to node-based Krea work.
2-3
tools Tony's rule calls for: not a Swiss army knife, a couple of tools that actually fit your workflow's steps.

The judgment · 8:00 PM

AI amplifies the craft you already have, not the one you don't.

Dom's boundary line, in response to a question about scope creep: designers are being asked to do more, but designers are also skilled in a profession that is not copywriting. AI can draft passable copy for anyone, but a trained copywriter will pull better copy out of it than a designer ever will, and a trained designer will pull better design out of it than anyone else. The point of prompting is learning to articulate your own expertise well enough that the machine gives it back amplified.

The same argument showed up as a caution, not just a boast: Dom cited Delta's 2018 facial-recognition study, where accuracy for Black women sat lowest of any group even after retraining, a persistent blind spot the data never fully corrected. His point was not that AI is broken. It is that engineers alone can't be the only ones building it. Designers carry culture, diversity, and lived experience into the room, and that is the actual argument for why designers lead.

58%
Delta's cited 2018 recognition rate for Black women, the lowest of any group tested, even after retraining.
1
craft to stay in: your own. AI raises the ceiling of the discipline you are trained in, it does not hand you someone else's.

Methods and prompts

Five methods to take with you

METHOD 01 · TAUGHT 7:15 part of: go deep not wide

Name your two or three tools

Map one real repeated task, then defend the two or three tools that cover it, not a scattershot kit. This is a judgment call the reader makes first, before the AI weighs in.

Where it came fromTony's rule from the tools walkthrough: the semester asks you to go deep on two or three tools that fit the steps of your own workflow, not shallow on a hundred. Dom added the reason: if everyone uses the same prompt libraries and the same kit, everyone gets the same output.Use it whenYou are about to add another AI tool and cannot say which repeated task in your week it actually covers, or you want to defend the tools you already pay for.

Working prompt

Here are the repeated steps in my workflow: [list them]. I answer first, then you check me: I've picked these two or three tools to cover it: [tools], and here's why each fits: [your reasoning]. Now tell me which step is still uncovered, and whether any tool I picked is redundant.

You will know it worked whenit names the exact step your list didn't cover or flags one specific tool as redundant, not a vague thumbs up.

METHOD 02 · TAUGHT ALL NIGHT part of: manage AI tone

Set a memory or tone rule

Tony's own move: he told Claude not to comment on his feelings unless he asks first. One sentence, saved once, instead of correcting the same drift every session.

Where it came fromTony showed the room a small habit from his own practice: instead of correcting an assistant's tone every session, he gave Claude one standing rule and had it saved to memory.Use it whenAn assistant keeps doing something you did not ask for, in the same way, in every new chat.

Working prompt

Update your memory with this rule: [state the tone or behavior you want changed, in one sentence]. Confirm you've saved it, then show me how you'd apply it the next time I ask you something that would normally trigger the old behavior.

You will know it worked whenit confirms the rule is saved, then walks through a next-time example that shows the new behavior instead of the old one.

METHOD 03 · TAUGHT 8:00 part of: keep the judgment human

Judge AI output with your own trained eye

Run a piece of AI-drafted copy or design work back through your own trained eye before deciding what stays. The reader judges first, the AI checks second.

Where it came fromDom's answer to a student worried about scope creep, being asked to become a copywriter too: don't. AI drafts for everyone, but your training decides what you can get out of it and what you can judge. The point of prompting is to ask for what your own training makes you best equipped to judge.Use it whenYou have an AI draft of real work in front of you and need to decide what stays, without letting the tool flatten your own eye.
Same idea elsewhereThe sea-of-sameness check (class 2) · Pick the relationship on purpose (class 2) · Make it fact-check itself before you read it (class 3)

Working prompt

Here's an AI-drafted pass on a real piece of work: [paste]. I answer first, then you check me: the lines I'd mark as only a trained designer's eye would catch are [your marks]. Now tell me honestly which of my marks a generalist could have made just as well, and where none of mine hold up.

You will know it worked whenit tells you plainly which of your marks a generalist could also have made, not just praises your eye across the board.

METHOD 04 · TAUGHT 8:00 part of: designers must lead

Audit for the cultural blind spot

Before trusting an AI output involving people, identity, or culture, name the one dominant assumption it defaults to, then have the model scan for what a narrower training set would miss.

Where it came fromDom's caution, not a boast: Delta's 2018 facial-recognition study, where accuracy for Black women stayed lowest even after retraining. His argument for why designers lead is that they carry culture and lived experience into a room that engineers alone cannot.Use it whenAn AI output involves people, identity, or culture and you are about to trust it.

Working prompt

Here is an AI output involving [a person, group, or cultural context]. I answer first, then you check me: the assumption I think it defaulted to is [your read]. Now scan for other blind spots a narrow or biased dataset could have caused, and be specific about who is underserved.

You will know it worked whenit names at least one blind spot beyond your own guess and says specifically who gets left out by it.

METHOD 05 · TAUGHT 8:45 part of: build a legacy, not a chat log

Draft your own agent CV entry

Dom's speculative frame: future job applications may list trained AI agents alongside human skills. Draft the entry now, grounded in your actual body of work, not a generic pitch.

Where it came fromDom's close, a speculation grounded in something real: future job applications may list trained AI agents alongside human skills, the way a salary negotiation might one day include the cost of agents trained on your design language.Use it whenYou want a first honest sentence about what an AI trained on your work could and could not do, before someone else writes it for you.

Working prompt

Based on my actual design language and body of work: [describe it], draft three versions of how I'd describe a trained AI agent extending my style, as if for a future job application. Keep every claim grounded in something I've actually made, nothing generic.

You will know it worked wheneach of the three versions points to something specific you actually made, not a generic claim any designer could use.

The close · 8:45 PM

A roomful of designers and a roster of trained agents to come

Dom closed on a story rather than a slide, speculative but grounded in something that had already happened to him: a convincing impersonation text from Charles Pratt, the school's founder, dead for years, arriving with spot-on context and timing. Real fraud, not a hypothetical. He paired it with a stranger idea for where the field is heading: future salary negotiations may include the cost of agents trained on your design language, the way Zaha Hadid's studio still carries her language forward after her death. AI agent due diligence, security, and data privacy become the new background check. None of it is settled. All of it is coming during this semester.

Try this prompt

Quiz me on the course's three pillars and Dom's two-or-three-tools rule. Then give me a real workflow of mine and make me name the tools that cover it before you weigh in.

You will know it worked whenit asks about the two named ideas before doing anything else, then holds off on judging your tool list until you've written it out.

Related classes

Where this class connects

The shelf

Tools and references

Tools that night

  • ChatGPT Plus, required, $20/month
  • Krea Pro, required, free with a course code
  • Adobe Firefly, free via Pratt Adobe Creative Cloud SSO
  • Named as worth exploring: Google Gemini, Claude, Suno, HeyGen, Ideogram
  • Miro, the semester's shared info hub

Named in the room

  • Sharon Panelo's "Artful Intelligence" framework, previewed for her later class
  • Zaha Hadid's studio, the design-legacy analogy for trained AI agents
  • Delta's 2018 facial-recognition study, the bias case for why designers lead
  • Thirteen guest lecturers named for the semester's second half
← Start of the course CLASS 1 OF 24 Class 2 → Demystifying AI