Work
Designing ThinkBuddy
A system-level AI assistant built for macOS power users
ThinkBuddy was a macOS AI assistant designed for power users who needed speed, control, and clarity.
Instead of asking users to adapt to AI, ThinkBuddy made advanced AI workflows feel more natural, accessible, and understandable.
The product became Product Hunt Product of the Day, gained strong traction among power users, and attracted early investor interest, including a16z.
- Role: Product Designer
- Platform: macOS
- Scope: Product Design, UX, Design System, Iconography, Brand Identity
Links
- Website: https://thinkbuddy.ai
- X: https://x.com/thinkbuddyai
Highlights
- Product Hunt - Product of the Day
- Strong adoption among power users
- Immediate investor interest, including a16z
- Designed end-to-end across product experience, interaction systems, iconography, and brand identity
Instant Access
AI Without Context Switching
Quick Ask made AI instantly accessible from anywhere on macOS.
Launched with a keyboard shortcut, it allowed users to ask quick questions, use voice input, and either return to their workflow or continue in the main chat.
Beyond a Single Answer
Rethinking Model Choice
Remix allowed users to ask multiple AI models at once, compare their answers, and continue with the best combination.
Instead of forcing users to choose a single model, ThinkBuddy helped them focus on the quality of the final answer.
Designing for Every Input
Context Comes From Everywhere
ThinkBuddy supported voice, transcripts, screenshots, files, videos, clipboard content, custom instructions, and scheduled prompts.
The goal was to make it easy to bring any kind of context into a conversation without breaking the workflow.
When Conversations Become Documents
Beyond the Chat Box
For longer prompts and formatted content, ThinkBuddy turned the chat input into an expandable editor.
Users could paste styled content, write in markdown, edit structured text, and work with document-like inputs directly inside the conversation.
Making AI Understandable
Helping Users Make Better Decisions
ThinkBuddy translated complex AI concepts into visual, easy-to-understand interface patterns.
Model explanations, memory controls, response summaries, and quality indicators helped users understand what to choose and why.
Transparency & Control
Control Without Clutter
As the number of AI models grew, ThinkBuddy needed controls that helped users shape the product around their own workflow.
Users could organize, hide, reorder, and prioritize models based on how they actually worked.
Usage analytics made model activity and consumption visible, helping users understand how they were using AI over time.
Building the System
Visual Language & Identity
ThinkBuddy needed a visual language that felt native to macOS while still giving the product a distinct identity.
The brand identity was designed to make the product feel cohesive and recognizable in a rapidly growing AI market.
I designed ThinkBuddy's interface patterns, iconography, and brand identity around concepts that did not yet have familiar visual conventions.
The custom icon system helped make advanced AI workflows easier to scan, understand, and reuse across the interface.
Outcome
ThinkBuddy was more than a set of AI features.
It was an attempt to make advanced AI easier to access, easier to control, and easier to understand in everyday workflows.
Results
- Product Hunt Product of the Day
- Strong adoption among power users
- Early investor interest, including a16z
- End-to-end product, UX, iconography, and brand identity design
- Early exploration of AI patterns that later became common in mainstream products
ThinkBuddy was not designed to make AI more powerful. It was designed to make AI easier to use.