Role
Lead Designer
Responsibilities
Gravity Sketch — AI for VR and Desktop
Gravity sketch had no AI in the product. I shipped image-to-3D in VR and image-to-image on desktop with 8,200+ AI generations and ~5,000 models imported into active design work, and 70% of users came back for more.
Where AI belonged across two very different surfaces: a VR headset and a desktop app.
A direction to raise funds, before there was anything built to show.
An app that had grown faster than its own structure, so nothing new had an obvious home.
I mapped the entire app and ran workshops with power users to test the findings in-product. Every subsequent release had to answer one question: where does this sit, and can users find it?
Customer interviews · design consultants · community feedback
01
Community users
Beginners, hobbyists, and power users exploring the tool, with a clear opportunity to become customers over time.
02
Customers
Power users, or those growing into that role, who are driven by productivity, repeatable workflows, and faster outcomes.
Customers didn’t want AI to design for them; they wanted it to accelerate their own thinking.
Start modelling with less friction
image > 3D (create base shapes from sketches)
Communicate ideas more clearly with less effort
image > image (similar to Vizcom)
Set up environments and add textures with ease
prompt > image (outputs are .hdr and .pbr)
Built with the CEO, this map defines the interaction model:
The architecture map showed me users don't go hunting in panels, three entry points, one pattern.
This is where the project nearly died. Corporate policies blocked customer teams from touching our AI features.
I shipped anyway, on two routes: image-to-3D into community beta so we'd learn fast, and a Vizcom integration so enterprise designers could use a tool their procurement had already cleared without leaving our app.
8,200+ AI generations, ~5,000 models imported into active design work since launch (Apr 2025)
813 unique VR users adopted the feature, released through GS Labs (opt-in beta).
59% completed the full flow — image to imported 3D model in scene
70% returned, repeat users averaging ~10 sessions
Validated the AI direction → shaped the vision roadmap developed with the CEO
The lesson wasn't about the model. It was about where AI sits in the workflow. People didn't want the tool to design for them, they wanted help getting past the blank page. Once we understood that, the design questions got easy.
I found the procurement blocker in the room with the customer, not in research phase.
Align success metric with the CEO together with the vision.