An interactive demo
A 60-second demo

Users don't know what they need.

So stop asking them to articulate it.

Three takes on the same product onboarding. One conversational. One graphical. One that uses both. You're the new user — just watch.

Act 1 of 3

Onboarding with a conversational interface

FlowBoard — Welcome
👋
Hi! I'm your assistant. What would you like to do?
↑
↖ A blank box. No hints. No suggestions. Just "figure it out yourself."
Good for
Open-ended, exploratory tasks. Feels human. No interface to learn.
Bad for
Structured tasks. Users can't ask for what they don't know exists.
Act 2 of 3

Onboarding with a graphical interface

FlowBoard — Get Started
Menu
Home
Projects
Tasks
Workspaces
Goals
Inbox
Reports
Choose a template to begin
📋 Sprint board
🎯 OKRs tracker
🐛 Bug triage
📅 Roadmap
📝 Meeting notes
🚀 Launch plan
Getting Started
Set up your workspace
Invite teammates
Connect integrations
↖ Now you can see everything. But which do you start with?
Good for
Structured, repeatable tasks. Surfaces capabilities. Predictable.
Bad for
First-time use. Shows the map but not the destination.
The framework

It depends on the task.

Conversational interface
When the task is ambiguous, exploratory, or generative
  • Lets users explore without a defined path
  • Handles open-ended, creative inputs well
  • Feels human and low-friction
Graphical interface
When the task is structured, repetitive, or high-stakes
  • Shows users what's possible at a glance
  • Guides them through clear, repeatable steps
  • Reduces effort for known actions
The question

What if we kept what conversation does well and added structure where it falls short?

Before
What would you like to do?
↑
+ structure
After
What would you like to do?
Plan a sprint
Onboard my team
Just exploring
↑
↖ Same conversation. But now the user can see what's possible.
Act 3 of 3

Onboarding with both, working together

FlowBoard — Welcome back
What brought you here today?
↑ Your turn — tap one
↖ One question. One tap. The product figured out what you needed.
The reveal

Why the combined approach worked

What just happened
1 AI inferred intent from one tap, not a typed prompt
2 The graphical interface reorganised itself around that intent
3 A single, specific next action replaced the menu of possibilities
↖ The graphical interface gave structure. The AI gave it intelligence. Neither could've done this alone.
Why this works
Users don't have to know what they need — the interface infers it. Structure handles affordance and predictability; AI handles personalisation and intent.
Users don't know what they need.
So stop asking them to articulate it.
Remember…
Do we have to talk about it?
From the talk by Max Keetley