Pencil: Building an Infinite Product Loop for Generating, Building, and Validating Product Hypotheses
These days, when I meet people in the industry, we often end up asking each other, "How are you using AI?" The ways people and companies use AI vary quite a bit, so these conversations are a good way to get a sense of how others are incorporating AI into their work. For me, that means quickly prototyping ideas with code, and sometimes taking those prototypes a step further by actually building and shipping UX experiments. Internally, we also have QA agents that identify issues and fix them directly, so having humans and AI work together has become a pretty natural part of how we work. Even though I use AI quite extensively, when I think about whether our organization has actually become as fast as I expected, I'm not sure the answer is yes. That's because using AI to make an individual task faster and helping an entire team move a product forward faster are two different problems. To make the whole team faster, it's not enough to simply finish each experiment more quickly. What really matters is how quickly we can learn from the result and move on to the next experiment. Lately, I've come to realize that what's particularly important is building a loop where the entire team can accumulate product learnings quickly, and those learnings can immediately feed into the next experiment.
