Alexa for a day: What we learned as a voice assistant

1 min read

“Okay, I’m starting the Krasinski case.”

“Wait, how do you spell that?”

I’m in Lindsey’s office, rapidly scribbling in a notebook. I’ve volunteered to be her personal timekeeping assistant for a day, sitting in her office, just waiting for her to ask me to record her billable activity. When she does, I scribble down all the information I can, so she can later enter my notes into her billing system. It’s hard work—I’m standing in as Alexa, Amazon’s (mostly) beloved voice assistant, and now I feel like I know why she sometimes answers my questions with such attitude.

So why go through all the trouble? We were using an experimental technique called “Wizard of Oz” testing, or “fake it until you make it,” where instead of building software to test your idea, you deploy members of your team to stand “behind the curtain” and act like the software. Through this open-ended testing, we were learning how people would react to having a time-tracking assistant capture their activity. If it worked, it would save people time, recover time entries they didn’t always bother to capture, and be a lot more fun than scribbling on stickies or pulling up a screen on a computer.

At Tali, we practice continuous learning and at every step in our development we ask ourselves, “What’s the most important thing we need to learn this week?” and “What’s the least amount of effort we can spend learning it?” As a small team, it’s critical that we deploy our efforts as effectively as possible, so we build learning into every step of what we do.

Sometimes we learn by prototyping our software, writing code we aren’t afraid to change or throw away, and sometimes we find an even more lightweight way to test an idea.

Instead of carrying out a lot of planning up front before we build our product, we maintain a long-term vision for our product, which guides us on where we want to be. We use small, inexpensive experiments to get there. In their book, Sense & Respond, Jeff Gothelf and Josh Seiden make a fantastic case for this approach to product development.

Continuous learning through experimentation keeps us close to the needs of our customers and to things we can build that are solutions.

Week after week, Tali gets better at making the chore of time tracking disappear, and with each step, we get closer to our vision: Stop tracking your time; start understanding it.