Long story short
For 7 years I was the only designer on UI Bakery - and I watched it become three different products. It started as a theme tool, turned into a no-code builder, then a low-code platform, and finally an AI-powered IDE where you build an internal app just by chatting with an agent. Through every pivot I owned the whole surface: UX architecture, design systems, the builder, database UX, org management, and the AI app-generation experience. The payoff was a bootstrapped SaaS that turned profitable and now runs inside hundreds of SMB and enterprise teams.
Problem
Low-code was supposed to be simpler, but building internal tools still meant wrestling fragmented builder, data, and workflow tools - users had to think like developers, and time-to-value dragged.
Solution
As the only designer across 3 major pivots, I rebuilt the whole thing end to end - drag-and-drop builder, real-time theming, a code-native IDE, a no-SQL internal database, org and roles, and an AI Agent that spins up a full app from one prompt.
Result
Grew to $1M+ ARR bootstrapped with thousands of users - up to 15× faster time-to-release, ~2-min AI app generation, and 3× fewer people per project across 10+ launches.
Context
UI Bakery never sat still. It went from a simple theme generator to a platform where you can build and customize apps with real code-level control - or just describe what you want to an AI Agent and watch it appear. Each step was a different product to design.
2019
Theme generator
Eva DS, Eva colors, Nebular (I also was a key designer on those products) dashboard theming tool.
2020
Website & app builder
No-code (with low-code elements) builder for apps & websites.
2022
Low-code platform for internal tools
Pivoted to a low-code platform for internal tools. This required a complete redesign of workflows, drag-and-drop UX, IDE experience, and integrations.
2023–2024
Low-code platform with AI-powered low-code IDE
Same system, but with AI chat-based in-code debugger, documentation, and coding assistant. Check PH release.
2025
UI Bakery Agent for internal app generation
Shift towards an AI-powered app generator, again reshaping the product experience. Check PH release.
Read more
By 2024, UI Bakery had grown into a $1M+ ARR business with no outside funding - hundreds of SMB and enterprise clients, thousands of daily users. As the solo Product Designer, I ran everything from research to the final pixel, pushing the product to feel more consistent, more developer-friendly, and ready for enterprise teams.
! Problem
It was sold as low-code, but building an internal tool still took real effort - data, UI, and logic all had to be wired up by hand.
Low-code cut the coding, not the complexity - users still had to think like developers.
Builder, data, and workflows lived in separate corners, so the experience felt disjointed.
All that setup and mental overhead made onboarding slow and pushed time-to-value out.
The new flow combined all use cases into one ecosystem covering app creation and lifecycle.
Design challenges
Seven-plus years as the only designer meant the hard parts kept stacking on top of each other:
Product evolution & pivots
3 pivots in a row: theme → no-code → low-code → AI
Turning each strategy shift into shipped UX
Solo ownership & speed
The only designer, owning it end to end
Moving fast without letting quality slide
Scaling complexity
Rebuilding flows and systems with every pivot
Going head-to-head with Retool, Appsmith, Replit, and the new AI tools
Solutions
What was done and what benefits were achieved for each solution:
Drag-and-drop Builder
I rebuilt the builder for faster, clearer layout work - tuning grid behavior, simplifying the logic, and giving it a structured sidebar.
People started building real interfaces in minutes, with a far gentler learning curve. The new grid (inspired by CSS Grid and React Grid Layout) doubled time-to-first-value.
Theming & Design Systems
I designed the UX for a scalable design-system and theming framework - users could tweak design parameters live, right on their running apps.
Themes could be swapped in seconds, no custom development needed.
AI Assistant
A chat-based in-code debugger, documentation helper, and coding assistant.
Adopted into users' day-to-day debugging inside the IDE, not a side feature.
Internal Database, Data Source library + Data Storage
I designed a native, Supabase-style database where users create and edit tables visually - no SQL - plus the Data Source library and a smooth flow for connecting external databases and APIs.
It became a core feature used by nearly every enterprise customer, and has needed no structural redesign since launch.
Organization
I built the multi-project structure from scratch - roles, permissions, environments, onboarding.
The early-2025 redesign landed well and visibly improved usability. Each iteration nudged onboarding conversion up another 2-4%.
UI Bakery Agent
The latest pivot brought an AI Agent that builds internal apps from a single prompt - generating UI, workflows, and data models on its own. I designed the whole AI experience, from app creation and debugging to motion-based feedback.
The pivot massively accelerated conversion (>10×).
Learn more
Research, Discovery & Validation
Since the project lasted many years and had many internal apps and parts, I used a multi-layer research and testing process:
Continuous feedback & metrics
I pulled signals from support, account managers, and in-product analytics to catch pain points early.
User research & testing
Surveys, CustDevs, and usability tests - corridor and A/B - to pressure-test decisions.
Behavioral analysis
FullStory and PostHog to replay sessions and map where friction met effort.
Impact & Results
Business outcomes
$1M+ ARR (bootstrapped, still growing)
Higher activation than traditional low-code
Thousands of SMB & enterprise users
3× fewer people per project
Strong post-activation retention
The AI Agent drove the speed step-change — up to 15× faster time-to-release and ~2-min app generation (full numbers in the AI Agent case)
Exact numbers are under NDA; these are directional.
Design leadership impact
Led end-to-end product design across the platform & AI Agent
Built and scaled 2 design systems that cut rework and sped up iteration
Tighter product consistency through a unified design system
Higher onboarding conversion across iterations
Held quality at speed (~72% SUS, confirmed by ongoing monitoring of users)
Solo-led every pivot, 10+ launches
Reflections and Takeaways
Pivoting solo, the design system stopped being a nice-to-have - it was the only way one person could keep three products coherent through every rewrite.
Designing for developers taught me to respect the code underneath: good low-code UX hides the complexity without pretending it isn't there.
The AI shift moved my job from drawing screens to shaping how a user's prompt becomes a working app - intent in, product out.
Every pivot rewarded shipping the smallest usable version fast and letting real usage, not my assumptions, decide what came next.
Reflection
Constant pivots and developer-driven workflows forced me to get faster, more structured, and more outcome-focused.
This project rewired how I think about product design: the value isn’t in a polished idea, it’s in getting something usable into people’s hands fast. Strong UX isn’t just interfaces - it’s behavior, timing, and trust across the whole experience.
