Design notes
Conversational AI Design
Technology is complicated. Conversations shouldn't be. The next generation of software will not be navigated through menus. It will be talked through.
The interface should learn the person
For decades, we have asked humans to learn software. We hand them manuals, onboarding tours, and dashboards dense with buttons. Conversational AI design flips that contract: the software should learn the person.
A menu is a list of everything the software will allow. It exists because the software could never ask you what you wanted. That constraint is gone, and the menu is still here. The opportunity now is to build interfaces that understand intent before the user has to translate it into clicks.
Clarity is a feature
In conversational AI design, clarity is not decoration. It is the feature. A single well timed question can replace dozens of buttons. A product that explains itself removes the need for training. When the interface can ask "what are you trying to do?" instead of presenting every possible option, the experience becomes smaller, faster, and more human.
This is why I believe software should begin with human struggle, not technical capability. Great product ideas often begin with frustration. The designer's job is to locate that frustration and let the conversation dissolve it.
AI is not the product
AI is not the product. What AI allows humans to do is the product. Conversational AI design is not about adding a chatbot to an existing screen. It is about rethinking what the screen was for in the first place.
Your next SaaS product may not need a dashboard. Dashboards were designed when software could only wait for commands. Agents can participate. One third companion. Two thirds interactive canvas. The companion works out intent; the canvas is where the answer becomes something you can act on.
The biggest opportunity may be translation
The biggest AI opportunity may not be automation. It may be translation. Translating intent into action. Translating technology into business decisions. Translating what a developer just said into what a founder actually needs to know.
A nontechnical founder shouldn't need a technical cofounder simply to understand what their developers are doing. The right conversational interface sits between a founder and their technology. Not to replace developers, but to create clarity. The point isn't better answers. It's better questions.
Design from struggle, not from capability
Every product I have built started as a sentence that annoyed me. Usually: "that's just how it works." Conversational AI design gives us the chance to stop accepting that sentence. We can build software that asks what the person was trying to do, not what they clicked.
Start from the human problem. Reveal the technology second. The technology is the answer to the friction, not the reason for it.
What I am thinking about now
- What replaces navigation when software can ask?
- Whether personality changes comprehension, or only enjoyment.
- How a company operates when every employee has a translator.
A few principles to build around
Software should explain itself.
The interface should not make people learn the software.
A conversation can replace dozens of buttons.
Products should begin with human struggle.
AI dramatically expands who gets to become a builder.
If you can't explain it in a sentence, you don't understand it yet.