Vizcom
Created by industrial designers from NVIDIA and Honda, Vizcom is a design tool that turns sketches into professional renders, eliminating the technical skills and resources traditional rendering requires. With no product leadership, this vision was short of reaching enterprise users and investors.
I joined Vizcom as the Founding Product Designer to establish design operations, shape the core functionalities, and develop its design language. My efforts helped raise funding, grow the design team, and initiate several enterprise adoptions.
Landscape
Industrial designers across all industries rely on 3D rendering to make their ideas come to life. With a high barrier to entry, requiring powerful hardware and considerable time to master, this process was neither accessible nor affordable.
However, sketching is available to everyone. With the advent of diffusion models, the gap between an idea and its depiction had been reduced to seconds. Connecting sketches to these AI models, Vizcom brought effortless rendering to artists, designers, and students alike.
Rendering tools are powerful, but complex.
Challenge
There were no UX specialists involved in the first release of Vizcom. Built by industrial designers and ML engineers, it was more of a proof-of-concept rather than a market-ready solution.
Vizcom needed a stable foundation, a framework towards implementing new features, and a rebuild of what was already there. With me at the helm of design, the founding team was complete and ready for a new chapter.
The Vizcom website and UI before I had joined.
Discovery
Our industrial designer CEO, an industrial designer Product Expert, and I made up the product design team. We discussed how current tools use AI, how much they rely on it, and how they market the fact. DALL-E was still the dominant solution, and Midjourney had just launched.
I sought to uncover what tools and devices users are coming from, how Vizcom fits into and shapes their workflow, and what upcoming technology we need to be ready for. Within days, I was ideating on viable directions.
To restructure the floating elements obstructing the canvas, I proposed three directions. We went with the sidebar, as it separated the UI from the canvas most effectively, eliminating distractions while drawing.
It also reflected the order of operations the best. Drawing happens first, and drawing tools are housed at the top. A textual prompt describing the drawn object is typed in, render parameters are set, and Vizcom is ready to go.
Our Discord community was an important extension of the team.
Thousands of designers and students kept a steady stream of feedback coming in. Validation was continuous, and there was less need for structured testing. It also made the UX curve easy to monitor.
I held semi-structured interviews with our founders to connect diverging ideas into concrete roadmaps. While individual designers formed the majority of our user base, I wanted us to prioritise the minority that actually pays us — Enterprise users.
To most users, AI tools still carried a lot of uncertainty with them. This resource from 2018 informed a lot of our thinking.
Information architecture
Positioning the sidebar to the right reflects the importance of the canvas and encourages left-to-right order of operation.
To preserve canvas space and minimise visual distractions, I held that we shouldn't use dual sidebars. Secondary features such as Layers, that didn't have to be visible at all times, could utilise sidebar tabs instead.
Most Vizcom users use a PC + drawing tablet. To prevent unintended sidebar taps on iPad, we would disable the sidebar when the user is drawing, in addition to built-in palm cancellation.
The initial version of Vizcom displayed a render preview while the user was drawing. I proposed phasing it out, as the previews used significant resources in the background, despite being low-quality and of questionable utility.
The sketch input vs. the render output.
Some iterations later, we introduced a tab bar for multi-file support. Put together, the key features; the home button, drawing tools, and the Generate button; were all housed in the corners, easily discoverable.
Instead of displaying the finished render in a small window, I suggested blowing it up to fit the screen by default. This didn't require additional resources, and allowed users to see their drawings transformed in real time.
Prototyping
Due to the precision and granularity of drawing as an input and the infinite variability of textual prompts, we couldn't validate design decisions without a fully functional version.
Aside from the workflows, an important component that needed testing were the drawing tools themselves — brushes, shapes, masks, and their appropriate settings.
I was handing off medium-fidelity designs directly to Engineering. The prototypes were then presented to select customers and community members, including designers from Autodesk, EA, and Lego.
The community involvement in Vizcom put the product in perpetual testing, with constant iterative loops between Engineering, the product expert, and me.
Working fast also allowed us to showcase these prototypes in sales pitches and investor meetings.
Design language
Our users were coming from Procreate and Photoshop, and the conventions of those tools informed our design language. Considering the lacking usability of Photoshop, though, we had to put them through rigorous iteration.
One notoriously flawed feature of design tools is the file explorer. Information overload with no visual distinction between files seems to be the convention. Wanting Vizcom to do better, I highlighted negative space, visual hierarchy, and something called information scent.
Project hierarchy is strict, files are grouped by status, and the sidebar
mirrors the main window.
I encouraged showcasing Vizcom renders across all media and all industries our users represented — be it automotive, furniture, fashion, or consumer tech.
The graphics are focused on a singular subject before a dark, neutral background. These are further overlaid with our hues and gradients, understated copy, and clean typography.
Validation
The issue with continuous and self-initiated validation from our community was that we weren't necessarily getting feedback on the features that needed it the most.
This became apparent when we introduced a feature called Inpainting, and built a new view around it — the Edit mode.
The default view in Vizcom.
Inpainting, as opposed to generating an entire image, modifies a specified area of it — replacing, removing, or adding objects. The inverse Outpainting feature generates objects outside the canvas, expanding it further into space.
These features require a rendered image on the canvas, so we housed them in a separate workspace we called Edit mode. After rendering, Edit mode would appear on the toolbar. Inside of it, we designed 6 new tools that would replace the default tool set and carry the new functionalities.
However, introducing 6 new tools and a new view to house them proved to only add friction.
Through reorganising the toolbar, we could cover the same functionality with a subset of the Mask tool and without a separate Edit Mode. This also meant my custom icon design efforts, though effective, were misguided.
Planned
Our stakeholders were pushing collaboration as a feature, allowing multiple users in one file. Our product expert and I weren't receptive of the idea, as it didn't reflect designer needs and seemed more like a trend than a useful feature.
A more agreeable vision was developing 3D capabilities, generating exportable 3D models from our 2D renders. Another feature we worked on were Palettes — custom AI models for Enterprise users to train on their own designs.
Delivery
We maintained a considerable overlap between Design and Engineering. I consistently consulted developers throughout the process, supported them hands-on in implementation, and carried out quality assurance duties.
We developed all features rapidly, without high-fidelity designs, a design system, or any extra weight slowing us down.
I established Vizcom's first design process and laid out a framework for future design work — raising the bar for design quality in a pre-Seed startup.
My design leadership also extended to brand identity, graphic design, social media, and web design. I concluded my position at Vizcom with a new website made in Webflow.
Result
With the new release of Vizcom, we saw significant adoption from Enterprise users across fashion and automotive design. The overall user base hit 100,000 designers, and the Vizcom team grew from 5 to 12 — including 2 new designers. And in early 2023, we raised $5 million in our Seed funding round.
As of October 2025, Vizcom have raised another $20 million in Series A, and $27 million in Series B. (source)