Most SaaS products greet new users with a wizard. Five screens of dropdowns, toggles, and text fields asking you to configure something you don't understand yet. The assumption is clear: the product needs your input before it can do anything useful.
That assumption made sense when software was a blank canvas. You had to tell the tool what you wanted because the tool had no way to figure it out. But we're past that now.
The onboarding wizard is a confession
When a product shows you an onboarding wizard, it's admitting something: “We don't know what you need, so you'll have to tell us.” Every dropdown is a decision the product is pushing onto you. Every toggle is a feature the team couldn't decide on, so they made it configurable.
The result: users spend their first five minutes answering questions instead of getting value. Most of those answers are wrong anyway — how would you know the right project template before you've used the product?
What AI-native onboarding looks like
We built Lova's onboarding as a conversation. You don't pick a template. You don't configure columns. You don't set up integrations. You just describe your project.
“We're building a mobile app for restaurant reservations. Three developers, one designer, launching in eight weeks.”
From that single message, Lova creates a board with columns that match your workflow, tasks that reflect your actual scope, and priorities based on your timeline. No wizard. No template gallery. No configuration at all.
The key insight: onboarding isthe product. The conversation that sets up your project is the same conversation you'll use to manage it. There's no separate “setup” phase. You start working immediately.
Why this matters for retention
Traditional onboarding has a dirty secret: most users never finish it. Industry benchmarks show that 40–60% of users who start a free trial never complete setup. They hit the wizard, feel overwhelmed, and close the tab.
Conversation-based onboarding inverts the dropout curve. Instead of front-loading complexity, it front-loads value. Your first interaction produces something useful — a structured project board — and every subsequent message refines it. The setup cost is zero. The switching cost builds with every conversation.
The template trap
Templates seem like they solve the onboarding problem. “Just pick the one that matches your use case.” But templates have two failure modes:
- Too generic.A “Software Project” template gives you To Do, In Progress, Done. That's not project management — that's a three-column spreadsheet.
- Too specific.A “Scrum Sprint Board” template assumes you run sprints, have a backlog, do retrospectives. If you don't, the template creates more confusion than it solves.
AI-generated boards don't have this problem. They're specific to your project because they're generated from your description of your project. No template library can match the specificity of “tell me what you're building and I'll set it up for you.”
The compounding advantage
Here's what makes this approach defensible: every conversation makes the product better at onboarding the next user. Not through a recommendation engine or collaborative filtering — through better AI. As language models improve, the onboarding improves. As we learn which board structures work for which project types, the AI's first suggestion gets closer to the user's ideal setup.
Traditional onboarding is frozen in code. AI-native onboarding gets better every month without shipping a single line of product code.
What we learned
Three things surprised us when we shipped conversation-based onboarding:
- Users tell you more than wizards extract.A text field that says “Describe your project” gets richer input than five dropdown menus ever will. People naturally include context, constraints, and priorities when they write freely.
- Refinement beats configuration.Users would rather say “actually, add a column for QA” after seeing the board than pre-configure a QA column before seeing anything. Show first, adjust second.
- The first AI message is the product pitch. When Lova responds with a structured board and a brief explanation of why it chose that structure, users immediately understand the value proposition. The onboarding is the demo.
Every product will eventually move to this model. The ones that wait will be stuck explaining to users why they still need a wizard when the competition just asks “what are you working on?”