AI is easy and you will be OK (some 2026 predictions)
AI is happening. Agents are happening. And you will be ok.
2025 was the year of Claude Code. Around the fall, the models (Anthropic, Gemini and openAI) got very good at writing code. (Yes you still need engineers. But also yes, 10x faster isn't unusual). Boris @ Anthropic, who created Claude Code, is an incredible engineer and has completely stopped writing code "by hand". Think about that.
What does that mean for product and UX people? A few thoughts:
First, yes, code is easier to automate, because it's easier to test. It's much harder to test if a product feature or UX design is "good". But don't get too comfy.
Our product and UX processes, tools and artifacts (Figma!) have been created in a world where writing code was expensive and slow. That world is gone. As engineering changes, product and UX will change too.
Gemini 3 right now is the model that is best at visual design. (Claude is the best at code.) As I teach in the course, any model capability that is economically valuable will improve rapidly because models can be post-trained on that capability.
"Synthetic users" (asking the model to be your user) is a terrible idea, but synthetic user data (AI generated user behavior extrapolated from real user behavior) can work really well.
The world tends to rhyme. "Hand writing code" ~ "pushing pixels"?
It's a wild new world, with a strange new material that we've conjured into existence. But you'll be ok. Two skills that I believe are quickly becoming much more valuable:
Talking to people with different backgrounds.
Digging in and looking at the data (evals!).
Does that sound like you?