Hachi AI Cohort 18
Hachi is a free AI search engine that makes exploring information fun, engaging, and personal. Ask open-ended questions and get conversational answers from distinct AI personalities—each with their own perspective. Say goodbye to boring top-10 lists and hello to fresh insights, hidden gems, and playful chats that feel more like talking to friends.
Hachi AI Cohort 18
directly impacted users based on testing
Worked closely with the founders and a powerhouse ux team
Presented in front of 300+ attendees
Why we're excited
Working on Hachi AI was one of those rare projects where ambiguity, creative freedom, and real impact all collided. The mission was exciting: reimagine how people connect with AI in a way that feels flexible, personal, and genuinely fun. But getting there wasn’t easy—we faced major pivots, conflicting directions, and the challenge of balancing playfulness with usability. Through all of it, I grew so much—learning how to navigate the unknown, adapt fast, and bring delight into a functional, scalable product. Seeing our work come to life and the founders light up during our final presentation? Easily one of the coolest moments of the journey.
The how
01
02
03
.png)
THE BEGINNING | Our first 3 weeks
In our first few weeks with Hachi AI, we focused on setting a strong research foundation. The founders had a product idea and a rough sense of who their users might be—but they needed help shaping it. With Coach Hannah guiding us and our research lead, Erjing, keeping us on track, we created a Product Requirements Document (PRD) and jumped straight into research. We moved quickly—examining 10+ direct and indirect competitors, running a full UX audit, and gathering over 200 data points through interviews and surveys. We spoke with 15+ users (some familiar with AI chats, others totally new), and a clear pattern emerged: users aged 16–28 wanted conversations that felt flexible, fun, and personal—not stiff or robotic. That insight set the tone for everything that followed.
ideate, constrain, & test
With our research in hand, we landed on our core question: “How might we design Hachi to enable flexible conversations and personalized recommendations, making search fun and shareable?” Our first idea was to design a gamified, friend-like AI—a chat companion that could do more than just talk. But after sharing early findings, the founders made a big pivot. They wanted to focus entirely on the chat experience and move away from the “AI friend” concept. It was a curveball. We had to toss out a lot of early work and start fresh, but it pushed us to explore two exciting new directions: a clean, chat-focused UI built for flexibility, and a visual novel-inspired interface that made Hachi feel more alive. We constantly iterated—wireframes on wireframes—gathering feedback, syncing weekly with the Hachi team, and adapting fast. It felt like working at a startup: high energy, lots of change, and tons of learning along the way.
.png)
innovate, spec, & ship
In the final stretch, we ended up merging the best of both concepts: the flexibility and openness of a clean chat UI, paired with the playful charm of a visual novel-inspired interface. It felt like the perfect balance between usability and personality. We polished everything, built out a full design system, and created a high-fidelity prototype that brought our happy path to life. Presenting it to the Hachi team and seeing their excitement made it all worth it. We wrapped up by handing off the Figma file, packed with design decisions and ready for implementation. The whole project was a whirlwind—but one filled with growth, late-night breakthroughs, and some of the most rewarding teamwork I’ve experienced.
Impact & shipped
Text input and answer prompts - supports our Q&A flow, making the experience more intuitive and open-ended.
What is the future of AI native interface and how do we bridge the gap?
• Designing workspaces that morph based on context vs static documents
• Canvas adapts in real-time to your workflow - not just another doc editor
• Content self-organizes based on relationships vs manual folder structures
• Information flows between blocks naturally, like liquid vs rigid containers
• Elements reshape and connect based on AI understanding of your work
• Built for flow state - content appears where needed, then vanishes
• Designing workspaces that morph based on context vs static documents
• Canvas adapts in real-time to your workflow - not just another doc editor
• Content self-organizes based on relationships vs manual folder structures
• Information flows between blocks naturally, like liquid vs rigid containers
• Elements reshape and connect based on AI understanding of your work
• Built for flow state - content appears where needed, then vanishes
Share - allows users to share Hachi’s fun answers, increasing engagement and interaction.
Group chat - allows users to add multiple Hachis to a conversation and seamlessly switch between their answers to explore different perspectives.
Hachi Reacts -adds personality that users love while helping the AI learn and personalize each character over time.
Some of our team memories
First Meeting with the client! Talking about our PRD with Coach Hannah
Getting high-level thoughts/critique from Hachi Founder, Nancy! This was during our pivot
Demo Day with Cohort 18! Sharing our 10-week journey designing for Hachi AI to 300+ attendees!