ABOUT
Before Scheme was Scheme, it was ThoughtTree. ThoughtTree was broken up into 2 separate products: ThoughtLab and ThoughtSpace. One was aimed at managing templates and day-to-day workflows while the other was meant to be a place for users to organize their thoughts and make sense of large documents. When I joined in September 2025, the idea was to combine the two products and create Scheme: a research and note-taking tool that uses AI in assisting users in interacting with their documents.
ROLE
Product Designer: Develop personas, ideate through sketching and low fidelity wireframes, and produce high-fidelity mockups and prototypes to define and communicate design solutions.
Tools: Figma, FigJam, v0, Mixpanel
Skills: UI/UX, Composition, Interaction Design, Design Standards, Whiteboarding, Information Architecture, Prototyping, Typography, Product Design Thinking, WCAG Compliance, Competitive Analysis, Contextual Inquiry, Data Analysis: Quantitative and Qualitative, Journey Mapping, Card Sorting
Redesign
Nobody understood how ThoughtTree and ThoughtLab worked. During VC demos — which at this stage of the product served as our most direct window into usability — a consistent pattern emerged: people were confused by what the templates in ThoughtLab were supposed to do, unclear on how ThoughtSpace functioned, and couldn't see how the two products related to each other. The experience wasn't just unintuitive — it was fundamentally fragmented. Users were being asked to learn two mental models before they could get value out of one product.
For an early-stage product trying to build traction, that kind of first-impression friction was a critical problem. If people couldn't grasp the product in a guided demo, they had no chance of figuring it out on their own.
When I joined the team in September 2025, the decision had been made to unify ThoughtLab and ThoughtSpace into a single product — Scheme — reimagined as an AI-powered research and note-taking tool. My challenge was to help design a unified experience that made the product's value immediately clear, eliminated the confusion between two disconnected systems, and gave users one coherent way to interact with their documents.
THE PROBLEM
DISCOVERY
Before I could design a unified experience, I had to understand what I was unifying — and why the current structure wasn't working.
I started by sitting down with the founders to understand the original vision behind ThoughtLab and ThoughtSpace. What were these products supposed to do? What problem was each one solving, and where did the team believe they connected? I needed to understand the intent before I could evaluate the execution, because the gap between the two was where the real design opportunities would live.
From there, I went hands-on with both products. I worked through every flow, every screen, every interaction — not as a designer evaluating UI, but as a user trying to get something done. I mapped out the full feature landscape across both products, documenting what existed, where functionality overlapped, and where the two experiences contradicted each other. What I found was that ThoughtLab and ThoughtSpace weren't just confusing to users — they were conceptually tangled. Features that should have lived together were split across products, while other features were duplicated in ways that made it unclear which product you were supposed to use for what.
I also watched recordings of product demos to see the confusion firsthand. I wasn't just looking for what people didn't understand — I was looking for the specific moments where they lost the thread. Where did their faces change? Where did the questions start? Where did the presenter have to stop and explain something that should have been self-evident? Those moments became my map of what the unified product needed to fix.
Three key insights came out of this process.
The split between ThoughtLab and ThoughtSpace was creating a decision burden before users ever reached the product's core value — people were getting stuck on where to start instead of what to do.
Templates — which had been the entire foundation of ThoughtLab — weren't just poorly presented. They fundamentally didn't align with where the product needed to go. They were adding conceptual weight without delivering proportional value, and no amount of better UI was going to fix that. The right move was to cut them entirely and only revisit the concept if and when real user demand emerged.
Third, the document interaction model in ThoughtSpace was actually the strongest part of the experience, but it was buried under a product structure that obscured it. These insights reshaped the entire direction of Scheme. Instead of trying to merge two products feature-for-feature, we made a harder and ultimately better decision: strip away what wasn't working, put document interaction at the center of the experience, and build a product with a single clear entry point that didn't ask users to choose between two mental models before they could do anything useful.
With the insights from my discovery work in hand, I moved straight into Figma. We were an early-stage startup operating at a pace that didn't leave room for a slow, linear design process. There was no luxury of sketching for a week, presenting three directions, and iterating through rounds of feedback. The team needed designs, and engineering was ready to build.
So I started working through the new information architecture and core flows directly in Figma — exploring how a unified Scheme experience should be structured, what the entry point should feel like, and how users would move through the product now that templates were gone and document interaction was the central focus.
This phase was fast, iterative, and at times messy. I was exploring and producing simultaneously — testing ideas in high enough fidelity that engineering could start building while I was still refining. That speed was a double-edged sword. It meant we were constantly shipping, constantly learning, and constantly putting real product in front of people. But it also meant work was going into production before it was fully resolved. Screens I was still iterating on were already being built. Interactions I hadn't fully thought through were already live.
It taught me something I carry with me now: even in a fast-moving environment — especially in a fast-moving environment — it's worth taking the time to sketch and set boundaries before jumping into high-fidelity tools. Not because the process needs to be slow, but because once your work is in Figma, it looks ready. And when it looks ready, it ships — whether it is or not. Establishing clear gates between exploration and production-ready work is a discipline that protects design quality without slowing the team down. It's a mistake I only needed to make once.
SOLUTION
ThoughtTree became Scheme — and with it, the entire interaction model changed.
The original interface was cluttered with buttons and controls across the top of the screen. Every feature had been given equal visual weight, which meant nothing felt important and users had no clear sense of where to start or what to do next. For a product that was supposed to help people think clearly, the interface was doing the opposite.
We solved this in two key moves. First, we introduced an all-encompassing search bar that served a dual purpose — users could prompt the AI agent to take action on their behalf or search their board to find specific content. Instead of scanning a toolbar full of buttons and trying to figure out which one did what they needed, users had a single, familiar entry point. One place to go, whether you wanted to do something or find something. It dramatically reduced the cognitive load of the interface and put the AI capability front and center in a way that felt natural rather than bolted on.
Second, we replaced the top-heavy button layout with a cursor toolbar anchored to the bottom center of the screen. This kept essential tools accessible without competing with the content on the board. It was a deliberate shift in hierarchy — the user's work became the focal point of the interface, and the tools moved into a supporting role. The board was the product, not the toolbar.
We also introduced the ability for users to share their boards and collaborate in real time. This was an important strategic addition — it moved Scheme from being a solo research tool into something teams could use together, which expanded the product's value proposition and made it stickier in a way that mattered for growth.
Together, these changes accomplished what the ThoughtLab/ThoughtSpace split never could. Instead of asking users to navigate between two products and a wall of controls, Scheme gave them one board, one search bar, and a clean set of tools. The product finally matched the simplicity of what it was supposed to do — help people think.
