Abstract

Plandid is an AI-driven workplace communication and collaboration tool that I designed to help product managers increase their productivity. Its prominent features include advanced document search, context summarization, and trackable collaborative task flow. Plandid's interface includes a unique "billboard/kanban board" function on the right that enables PMs to manage important cross-functional initiatives in real-time, along with occasional context-triggered AI suggestions during chats. As an ambitious UX designer with an entrepreneurial spirit, I decided to leverage my thesis as an opportunity to gain a taste of operating a startup business. After months of effort, my team and I achieved several successes, including recruiting five dedicated talents with diverse backgrounds and experience, winning the B2B track of the startup bootcamp at NYU Entrepreneurial Institute, and being shortlisted for the final round of the Startup Sprint interview. Although originally planned as a web-based application, Plandid failed to take off due to unforeseen circumstances during the execution phase. A few weeks after our decision to discontinue the project, Slack announced the rollout of a new feature called Slack Canvas, which was a bitter-sweet moment for the team. While the high overlap between Slack Canvas and Plandid validated our ideas, it was also a strategic mistake to build a business in a saturated market with highly resourceful competitors. We lacked a proprietary advantage and were consumed before fruition. Having failed multiple times in my life, including the Plandid project, I understand that success is not a linear process. Each failure is a step closer to success. I hope this thesis project can serve as a cautionary tale for myself and fellow designers on what to do and not to do when building companies.

Research

We conducted 27 in-depth interviews, 15 mock-up review workshops, and a comprehensive competitor analysis. After research synthesis, we built an early adopter profile - the person will be a tech-savvy founder or early employee of a growth-phase startup whose day-to-day tasks involve agile team collaboration. Their needs include easy file sharing/preview/summary and advanced document search functions among the noises of millions of messages. Throughout the process, we also learned a few lessons on what we can do better next time: for example, there’s a diminishing return of in-depth interviews once we hit certain numbers; we could’ve built a screener to filter who we interview once we hit the evaluative research stage, competitive analysis already gave us insights on the fact that our angle to the market was a dangerous path, but we didn’t pick it out… We did a good job of gathering information, but we were too invested in the quantity rather than the quality of the information. So if given a second chance, I will take a step back once every while during the research, and organize insight synthesis a lot more frequently.

Technical Details

For the technical front, we were looking at the OpenAI’s GPT-4 API and calculating the cost of different use cases.