Writings

Understand user flows without friction

I built PathKit because I got tired of the same meeting.

You’re sharing Figma. Twelve screens, the happy path, left to right. The client is with you. It feels finished. Then they ask what happens if the card is declined. Or if they’re logged out. Or whether you can even reach this screen from that one. The designer starts scrolling. A page of artboards. Nobody in the room can hold the whole thing. The founder asks how big this is. The answer everyone has is twelve. The answer in the product is not.

I have never had a client who didn’t hit that wall. Not because they weren’t paying attention. Because the product got big. Complex UX is a lot of small decisions stacked until the flow is larger than a file of frames can explain. We always tried to understand it in Figma. Figma is the wrong place.

Figma is very good at drawings, and very good at a path somebody clicks. You tap through the story. It feels complete because you stayed on the path. The rest of the product is not in the room. Empty, offline, declined, expired, locked out. Hidden work is not visible. So the project always looks smaller than it is. Then you start building, and it isn’t.

PathKit does one thing. It puts a user flow on a canvas, holes included, so a room can review it, audit it, and actually see how big it is.

Getting it on the canvas

I built a Figma plugin that syncs your artboards over. If you’d rather not, drop the files on the canvas. That’s it.

The point was to make this easy. Not a setup. Not a new file format. You already have the screens. A minute later the whole room is looking at the flow instead of scrolling a page of frames.

The number I stopped ignoring

On one booking flow the split was twelve drawn, forty one if you ship it. I’m not saying that’s a study. I’m saying that’s what it looked like when I stopped treating the demo as the spec.

A client can follow twelve. They cannot follow forty one if forty one is a pile of frames. They nod in the review. A week later they’re lost, because the holes were never in front of them. Neither was the size of the thing.

The canvas

A screen nobody designed shows up as a ghost with the reason attached. You can point at it. A founder can point at it. A developer can say that’s two weeks. The PM can see the project is not the twelve-screen story they signed off on.

That’s the power of it. Not generating more screens. Making the hidden work visible, so the people who have to decide can decide while it’s still cheap.

The loop

The loop I was stuck in looked like this. Draw the happy path. Present it. Get a nod. Hand it off. The other states get designed later, alone, in a ticket, by whoever picked it up. The client never saw them. The founder never priced them. That’s a process failure, not a drawing failure.

An agent made that loop expensive. It will build exactly what you showed it. It has no way to know a card can be declined if you never drew declined. Fast, faithful, twelve screens. Then the project is the forty one nobody scoped.

The work was never drawing the twelve. It was deciding, together, which forty one there are.

Just the flow

PathKit can find holes and draw ghosts and turn a Figma page into a graph. But if you just want an easy place to drop a flow, sit with it, and see how big it actually is, it does that just fine too.

It’s at pathkit.ai.