Last week, you received our AI maturity model. Most companies we talk to are currently in Phase 2.
What does that mean? You've bought ChatGPT or Copilot licenses. A few people have their own assistants. Marketing occasionally runs automated workflows. So, "you're using AI."
But you're still just chatting with it. AI is a prettier Google. That's not using AI, that's a symptom of inertia.
Technologically and in terms of mindset, we're already elsewhere. And this is exactly what we went through in the spring with people in the AI Design Lab and AI Marketing Lab.
From Chat to Agents
Phase 2 is chat. You ask, the model answers. Even custom assistants are just more sophisticated chat. AI consults, people do the work.
Phase 3 is an agent. You work in a tool with an agent layer — in our case, it's Claude Code. You build your own agent, equip it with skills and data access, and it plans its steps, does the work, and saves the result. AI does the work, people check the outputs.
You evolve from a specialist into an agent manager.
Breakthrough Habits
Participants in our spring trainings stopped doing things manually and started building systems that work for them. Three habits are common to all:
- Agents instead of prompting — in Claude Code, you build an agent once and then just assign tasks and monitor it.
- Knowledge base for agents — structured context that the agent accesses itself and can work with.
- Deep research as a source of information — you ask one precise question, the model breaks it down into dozens of queries, goes through hundreds of sources, and returns a synthesis.
Participants developed a wide range of skills for this, from data sense-making to prototyping.
What do we see in both training sessions? The boundaries between disciplines are starting to blur. A designer can analyze customer behavior in an application themselves. A marketer can build a landing page. Both thanks to well-built agents. This is gradually changing the game we play.
A one-off AI course doesn't make sense
AI changes every quarter. And the pace will likely increase for several more years! A course recorded in January is half-dead by August. The tools we taught in spring have different capabilities today than they did three months ago.
That's why AI Design Lab and AI Marketing Lab run in cycles.
If you get the training sessions today:
- You will immediately receive recordings of the entire spring session.
- This autumn, you'll review everything live with our new insights.
- You'll gain access to recordings of every future cycle.
Start with one person. Get access for your most innovative marketer or designer now (maybe that's you!). Instead of just using better prompts, they'll start tasking agents. And that will gradually change everything for you.
The illiterate of the 21st century will not be those who cannot read and write, but those who cannot learn, unlearn, and relearn.
– Alvin Toffler
Weekend Reading
Great work isn't done by smarter people. It's done by those who refused to be bored.
