Onboarding in AI Products is Breaking
I've been thinking about onboarding in AI products for a while now. And one thing I've noticed is, most AI companies are heavily focused on features, models, and shipping the next shiny thing.
And don't get me wrong, that's a good milestone. It shows progress. But I feel like onboarding is the missing piece.
Aside from problem-solving itself, onboarding is one of the most important parts of the experience. Because at the end of the day, when you keep dropping features and I'm a user of your product, what happens when I leave for a bit and come back?
Let's say I come back after a week. All of a sudden, it feels like I'm using a completely different product. It becomes disorienting.
I remember using Cursor for a project. I got used to it, understood the flow, and everything made sense. Then I came back a few weeks later. Everything had changed.
I couldn't find certain things. Some icons didn't make sense anymore. I had to hover over elements just to understand what they did. I was lost.
And I've noticed this pattern across a lot of AI products. The care for onboarding is almost zero. Instead, there's a heavy reliance on documentation.
If you want to understand how to use the product, you either:
- Watch a video walkthrough
- Or go through their documentation That approach pushes the learning outside the product. Onboarding should live inside the experience. It should guide users as they use the product, not send them somewhere else to figure things out.
Another thing I've noticed is that many of these products assume you already have a certain level of knowledge before using them. So if you come in without that knowledge, navigating the product becomes difficult. And that's where the gap starts to show.
We also need to think about the different types of users we're designing for. It's not just "users" as one group.
There are:
- New users
- Existing users
And even within existing users:
- Active users
- Users who come back occasionally Each of these groups experiences the product differently. So the way we onboard them cannot be the same.
Now, even within new users, there are different types.
There are users who already have some level of knowledge or domain expertise. They understand how these tools generally work, so navigating the product becomes relatively easy for them. They can figure things out.
Then there are users who are completely new. They don't necessarily have domain expertise, but they've seen something:
- A video
- A post
- A demo And based on that, they believe,
"This product can solve this specific problem for me."
So they come in with a very clear expectation. They are focused on solving that one thing.
At that point, they're not trying to understand your entire product. They're trying to achieve a specific outcome. And if the onboarding doesn't guide them to that quickly, it becomes frustrating.
So when we say "new users," it's not one category.
It's people with different levels of context, different expectations, and different entry points into the product.
And that should directly influence how onboarding is designed.
One thing that stands out to me is how most AI products are being designed with power users in mind. And I don't fully understand why. Because not everyone is a power user.
For a new user, there are only certain features that matter at the beginning. They don't need to see everything. Right now, it feels like there's an attempt to give visibility to everything at once.
And that just creates friction.
Now, here's the tricky part. Sometimes, even with all this friction, people still use the product. And that can be misleading. Because in many cases, the product is solving a very real and critical problem.
So users are willing to go through the pain. They're trying to get to a result they care about.
And that's why you sometimes see products dominate for a while...Until a competitor comes in, simplifies the experience, and starts gaining traction.
Another important thing,
Onboarding is not just about the first-time experience. It's ongoing.
Every time you:
- Ship a new feature
- Change the layout
- Update how something works You are affecting the user's mental model. And when that mental model is disrupted, users need guidance to adjust.
I think this is where most products are currently failing. There's a lot of excitement around shipping. But not enough care around helping users keep up.
As we move into a world where AI is making it easier to build and ship faster, I think we need to slow down just a bit. We don't need to slow innovation, we need to improve the experience around it.