Many career changers, students, and recent graduates struggle to find career opportunities, while businesses often lack the time and resources to source fresh talent. Kaie solves this by acting as an agentic AI career partner: it collects explorers’ skills, goals, and preferences through a conversational onboarding flow, then autonomously matches them to businesses and facilitates introductions over email. As a Product Designer at Kaie, I designed the end-to-end onboarding and profile experience that powers this matching system: from the AI chat-based signup to structured profile creation, ensuring explorers are fully ‘match-ready.’ My role included user experience research, ideation, mid- and hi-fi design, prototyping, and close collaboration with developers to translate these flows into development while aligning with the founders’ business goals.
Company:
Kaie
My Role:
Product Designer, UX Researcher
Year:
2025
Tools Used:
Figma, FigJam, Miro, Notion
Kaie isn’t a traditional job board. The product goal was to make early-career opportunity discovery feel less transactional and more supportive, while still gathering the data required for accurate matching. The challenge was onboarding: we needed high-quality, high-signal profile information, but asking too much upfront increased fatigue and drop-off. I worked on the explorer side of the experience, owning research, flow design, UI, and handoff for onboarding and profile creation in collaboration with the founders, engineers, and another product designer.
We initially explored a marketplace-style dashboard where users could browse opportunities and express interest, similar to a job board. Through early feedback and product direction, we realized this model pulled Kaie toward the same stressful “apply and wait” behavior we were trying to avoid. Instead, we shifted toward an agentic approach: Kaie would do the matching once we had the right inputs, and users would gain control through a clear profile they could review and edit after onboarding.
To validate the direction, we ran and synthesized usability testing (around 15 sessions) across two onboarding approaches: a conversational, chat-based flow and a more traditional form flow. Chat felt more human and trust-building, but became tiring when it tried to collect everything. Forms felt faster and familiar, but less personal. Across both, users consistently wanted clear progress, fewer questions up front, and reassurance they could edit later. The key insight was that onboarding was doing two different jobs at once, and needed to be split.
The final solution was a two-part onboarding flow. Part 1 used a lightweight conversational experience to create an account, build trust, and capture only essential information. Part 2 moved users into a structured, multi-step profile builder designed to collect higher-signal data for matching, with clear grouping, progress tracking, and editability baked in. This separation reduced perceived length and made the experience feel more intentional: “sign up quickly now, finish your profile with confidence next.”
After shipping, onboarding completion improved by 22% (based on the team’s tracking), and the flow established a scalable foundation for future iteration on matching quality and introduction moments.
