One of my co-founders, Han, started many mornings with a cup of coffee and an inbox full of customer emails. However, he didn’t always have the time to meticulously craft email replies. That’s when the idea of using artificial intelligence to automate email responses first materialized.
Today, my team and I launched Promptly, a CRM tool that uses company knowledge bases to instantly automate replies to customer emails. Despite many sleepless nights and bottles of sweet tea, we genuinely couldn’t have launched on Product Hunt without “The Office”.
Where does “The Office” come in?
We were committed to the idea, the research, and the technology. But our ideation process sorely lacked feedback from our target user base. As college students, companies with fully functioning customer support teams were beyond our personal networks. We found ourselves clumsily guessing at what our users needed.
Switching gears, we agreed to design and build a minimum viable product (MVP) in just two weeks. The goal of the MVP was to quickly get user feedback on the product direction of our tool.
On a whim, I suggested that we train the AI model using information about Dunder Mifflin from “The Office”. After all, the paper company does boast a stunning reputation for customer support. The demo was designed for users to ask questions as if they were communicating with Dunder Mifflin’s customer service reps themselves. With just a landing page and the interactive demo, we launched Dunder Email on Product Hunt.
A shot in the dark
To our surprise, Dunder Email finished as the #5 Product of the Day, and 1000+ users interacted with the demo. In retrospect, I theorize that using “The Office” made an unfamiliar technology feel much more approachable.
And Product Hunt certainly was not shy. My team and I read through every single demo submission and had quite a few laughs. We received questions ranging from “When is Pretzel Day?” to “Can I get Pam’s phone number?” to “Who is the Scranton Strangler?” In all honesty, our system was no match for many of the questions, but it wasn’t a waste of time by any means. Through the demo, we improved how the AI model leverages the knowledge base and identified patterns among the questions asked.
As the product gained traction, companies also reached out to share their thoughts on how our technology could fit into their support teams. The conversations following the Product Hunt launch were far more valuable than the number of upvotes or comments we received.
Back to the drawing board
Dunder Email gave us an invaluable foundation to build off of. We brainstormed new ways to address our users’ needs regarding internal documentation and data privacy. Promptly emerged on the other side with the following features:
- Automatically draft email replies within Gmail
- Update the knowledge base through the Promptly dashboard
- Organize emails by folders and tabs
Pre-launch, we dedicated many hours to create the website, promotional video, marketing graphics, and video demos — all in hopes of getting feedback when it came time to launch. We recognized that the process for creating a product is fluid and (sometimes painstakingly!) iterative.
Lessons learned
Make assumptions…but test them. We had a vision of a reimagined customer service experience. It’s clear that we didn’t follow conventional routes of user research and brainstorming. Instead, we placed our bets on the Dunder Mifflin demo even if it appeared like a humorous joke to some. If the initial product hadn’t received a positive response, we were ready to scrap everything and pivot. MVPs are extremely lightweight so there are no hard feelings.
Get feedback whenever and wherever. It felt tempting to pigeonhole our product since it was specific to the customer service space. When you’re knee-deep in a project, it’s easy to make common-sense mistakes. Do what you can to get a pair of fresh eyes on your product, even if it's your friends or family. Building the MVP and leveraging the “The Office” made it easy to share the product for feedback. There’s value in shipping early and often, and feedback helps move the process along.
Be open to conversations of all kinds. As a team, we sat in on countless Zoom calls to talk about Promptly. We met with friends, mentors, companies, “The Office” enthusiasts, and even competitors. There wasn’t one defining call that made everything click into place. In fact, most of these calls didn’t lead to anything substantial. However, with each conversation, we narrowed our people problem and our own perception of the product became clearer. My co-founders and I made many vital product decisions in post-call huddles. There’s no better way to move past a roadblock than to talk through it.
And finally, be brave. For both Promptly and Dunder Email, I felt intimidated by the launch. I was put in a position where my work would be critiqued. Yes, it’s vulnerable…but it’s also quite sincere. Don’t feel defined by the number of upvotes or the ranking of your post. You’re given an opportunity to reflect and improve upon your product — it’s a wonderful realization that what you put into the world is always malleable. And let’s be real, it can’t be worse than Sabre printers spontaneously combusting.
Despite a long and tiresome journey to launch, I’m energized by the direction that Promptly is headed in. You can support Promptly on Product Hunt or learn more at promptly.to!
Special thanks to my co-founders, Han and Maggie, for many work-session-turned-karaoke-nights and Product Hunt, for being a space where creators can share their visions.
Thanks for scrolling :-) Let’s link on Twitter or find me at reneehoh.github.io.
