Appsmith

When 42 Widgets Walk Into a Bar... and None of Them Match

Product Designer, 2022-2023

The Chaos!

Imagine you're building an internal app.
You drag a Form widget onto the canvas. Cool. Then a Text Input. Wait, why do these look like they're from different decades?
Add a Button. Oh god, now there's three design languages fighting for dominance.

With 42+ widgets all doing their own thing, the platform was like a music festival where every band showed up in a different genre. Developers just wanted to build internal tools quickly, but making them look even remotely professional? That was a whole side quest.

My mission

Bring order to the chaos. Make these widgets play nice together. Give developers a fighting chance in no time. Restoring trust for Org using internal app that matches to their brand

How I Fixed It

Chapter : 1

Became a Git Lurker (The Good Kind)

Appsmith's open source, which means real developers complaining in real time. I basically lived on GitHub, watching people hit the same walls over and over.

One dev literally said: "I just want my table to not look like a spreadsheet from 2003."

Fair Point.

So we documented everything in Figma, Notion and Jira.
Tracked patterns in Dovetail.
The message was loud and clear: developers wanted beauty AND brains. Not an either/or situation.

The Widget Intervention

Let's discuss Table /42 Widgets

The Table widget was public enemy number one. Most used. Most complained about. Most "why does this icon look like it time-traveled from Windows XP?"

The mess we inherited:

Icons that didn't match (like, at all)
Typography doing whatever it wanted
Spacing that made you question reality
Interactions so complicated they needed a manual
It was almost impossible to make them feel compliant with company's existing design language = Trust issue
Newly onboarded developers seeking support assistance soon after attamting to build their first app

The fix - EAT

Completely redesigning it? Too risky. This thing was everywhere. So we went with a strategic reskin, keeping the core behavior but making it look like it belonged in 2022.

Enumerate

Unified all the inconsistencies
Icons, Spacing, States. interaction based on report and pattern recognized

Address

Added three theme presets:
Default, Sharp, Round. Pick your vibe, Full customization from there with improved customization pane

Test

Made sure data binding from databases and APIs didn't break
(because that would've been a disaster)

Locked and Limited to Make It Yours

Why customization sucks?

Before, customization was a joke. You got corner radius. Shadow. Some predefined alignment options. That's it. Want to change the overall feel of your app? Want it to look sleek? Playful? Corporate? Too bad.

Developers were frustrated. Everyone was building apps that felt generic because there was literally no way to make them feel personal.

What We Built

We overhauled the customization panel. Instead of scattered, useless presets, designers got three powerful theme starting points that actually transformed how the entire app looked.

Default, Sharp, Rounded. Pick one. Make it yours.

You weren't locked in. Every setting was tweakable. And once you nailed your perfect combination? Save it as a custom theme to reuse instantly.

Why This Worked?

The Speed Plus Control Play : Designers needed two things: the ability to move fast AND the power to customize. We gave them both without breaking the codes.

Speed

The three presets handled speed. Pick Default, Sharp, or Rounded, and boom, your app instantly looks intentional instead of default. No more generic vibes.

Flexibility

The customization options handled control. Want darker shadows? Sharper corners? Different spacing? Tweak it. The custom theme feature locked it all in so you're never making those decisions twice.

Win

That combo is what made developers say "finally, I can build a design system that actually sticks around.
They went from frustrated to empowered. Apps started looking intentional instead of default.

Repete this for all 42 Widgets !

Why this approach slapped

Engineering team was busy with performance stuff, so low lift = win

Didn't nuke anyone's existing workflows

Gave us intel on what else developers actually needed

And we end-up with

47%

Drop in support request

+27%

New signups and collab

300+

Apps opts for upgrading to new widgets system upon release

60%

Increase in developer satisfaction score

What I Took Away

Ship the MVP, not the masterpiece

Perfect is the enemy of shipped. Get something out there, learn from real usage, iterate. Rinse and repeat.

User research is a lifestyle, not a phase

You don't "finish" learning from users. You keep listening, forever.

PM and Engineering is your best friend

Loop them in early. Understand what's possible, what's hard, what's "are you kidding me?" Then design accordingly. Saves you from the pain of redoing everything three times.