An exclusive FPD 2-week live course for designers, leads, and PMs ready to move past prompt-dependency.
For over twenty years, I have been teaching, consulting, and designing at the intersection of human behavior and emerging technology. I have watched interfaces evolve from the rigid precision of command lines to the fluid gestures of touch screens. And if there is one thing I have learned, it is this: the best interfaces disappear. You don't think about them. You just do things.
So here is what puzzles me about the current generation of AI products: the command line has come roaring back into the foreground, and not in a good way.
We have regressed. Except now the command parser gets moody and occasionally invents things.
This course is about what comes next.
What's wrong with most AI design content
Most AI design courses teach you to use AI tools. This one teaches you to design AI products. That distinction matters more than it sounds.
Tool fluency is not product design fluency. Knowing how to prompt Cursor or generate ten Midjourney variations is a Tuesday afternoon's worth of self-teaching. Designing AI products that people understand, trust, and choose to come back to — that is the work nobody is teaching well, and it is the work the field is failing at.
The data is clear:
- Roughly 85% of AI initiatives fail — and most of those failures are choice-of-application failures, not execution failures.
- AI features are launching faster than UX practices are adapting. Most teams are shipping into a vacuum.
- The senior designers I talk to across Pharma, Healthcare, Fintech, and Logistics tell me the same thing: we know how to ship deterministic solutions. We don't know how to ship probability.
The 7 Sins of AI Product Design
Before you can design the cure, you have to name the disease.
I teach a diagnostic frame for what goes wrong in AI products. Irrational Lab (San Francisco) coined it the 7 Sins of AI Product Design:
- Neglect — designing AI as an afterthought, bolted onto a product that did not need it
- Obscurity — building features users cannot find, cannot understand, or cannot trust
- Gluttony — eating data, attention, and screen real estate without giving proportional value
- Mediocrity — shipping outputs that are fine, when the alternative is doing nothing
- Vanity — designing for the demo, not the daily use
- Tyranny — removing user control in the name of "smart" defaults
- Envy — copying the dominant chat-box pattern when the product needed something else entirely
Most AI products you can name commit at least three of these. The good ones — the ones users choose to come back to — have systematically engineered the sins out.
This course teaches you how.
The 6 HAX Principles
The cure side of the work runs on six load-bearing beams. I call them the 6 HAX Principles:
- Empathy first — design for the user's real intent, not the AI's capabilities
- Automation vs. augmentation — choose deliberately, every time
- Transparency and confidence — calibrate trust, don't manufacture it
- Real control and editability — AI output should never feel final
- Graceful failure — AI fails differently than software fails; design for it
- Mental model shaping — people don't have a model for AI yet; you give them one
These are not invented from thin air. They are the synthesis of two decades of fieldwork at Optimizer.pt — across pharmaceutical platforms, hospital workflows, financial services, and logistics tooling — and a thorough reading of the public canon: Microsoft HAX Toolkit, Google's People + AI Guidebook, Apple HIG for ML, GitHub Copilot guidelines, the Shape of AI catalogue, AIverse.design, and the academic literature on human-AI interaction.
I take from all of them. The 6 principles and the 7 Sins are how I integrate them into something a senior designer can actually ship.
What you will leave with
Three concrete artifacts you can use the day the course ends:
- A 7 Sins audit of one AI product you ship or use daily — written in a format you can hand to your engineering and product peers without translation.
- A 6 HAX-principle redesign brief for one feature in your real work — concrete enough to scope, opinionated enough to defend.
- A portfolio-grade case study showing the before, the diagnosis, and the redesign — usable in interviews, internal reviews, and stakeholder meetings.
This is not a theory from an ivory tower. This is fieldwork.
Who this is for
- Senior product, UX, UI, and interaction designers working on, or about to work on, AI features
- PMs who need a shared vocabulary with their design teams for AI work
- Design leads and Heads of Design scoping AI initiatives across multiple teams
- UX researchers who want to design and evaluate AI products with rigor
Who this is not for
- People looking for a tool tutorial. Cursor and Midjourney change weekly. The principles do not.
- Beginners with no product experience. This course assumes you have shipped real software for real users.
- Those looking for hype. AI is a design medium, not a magic ingredient.
Across thirty-plus years of watching interfaces evolve from rigid command lines to direct-manipulation GUIs to gestural, voice, and ambient surfaces, I expected the AI wave to push us further along that arc. Instead, I watched it snap us back to the command line interface.
I would like your help to push it forward again.
About Christian Kuhn
Christian has over 20 years of international UX experience and currently leads the UX Center of Competence at Optimizer in Porto, Portugal. He drives the development of award-winning, user-centered product solutions through UX research and Behavior informed design. He contributes to Harvard University Extension School, Singularity University, Nomura Research Institute Japan, and many more. He is a published author on UX Design, AI Product Design and Behavior Design topics. He is teaching UX Leadership, Behavior Design and Human-Centered AI Experiences (HAX) at TheStarter.io. PortoAX Activist. Jury Member of the UX Nordic Award 2026.
FAQ
- Are sessions recorded? Yes. Live sessions are recorded and posted within 24 hours. Recordings stay available for 6 months.
- What time zone? Live sessions run 18:00–20:00 CET (CEST in summer). That is post-workday for everyone in Europe and mid-afternoon for the East Coast US.
- Do I need to know machine learning? No. You need to know how to ship products. The course is about design, research, and product decisions — not model training.
- Is there a certificate? Yes. All tiers receive a course certificate.
- What if I can't make a live session? Recordings cover you. Most students miss one or two — the cohort Slack and on-demand modules keep you on track.
- Will Christian actually answer my question? Teach assistants moderate the chat queue and surface the best questions. Christian answers 6–10 per session. There is no guarantee any specific question gets answered live, but the pod and Slack channels mean nothing important goes unanswered.
- Refund policy? 14-day refund window from Course start. Transfer to the next Course allowed up to 7 days before start date.
Schedule
Open Edition (4 online lessons x 2 hours)
Start: Monday, 9th of November – End: Wednesday, 18th of November
Every Monday and Wednesday from 18:00 to 20:00 (Copenhagen time).
Price
€ 390
Get Your Ticket
