New Deal for Pros (Alpha)
AI-generated memories | Thumbtack | 2025
The problem...
The product organization created a specific team, named "New Deal for Pros" and I was added as the content designer and prompt engineer. From H2 of 2025 through H1 of 2026, we are designing a 0-1 experience that reimagines how our product might work for pros.​
​
One of Thumbtack's main goals is to help pros get hired by homeowners. During onboarding, pros have always had to laboriously enroll in each type of job they're willing to take. These are called "job preferences." The process is not only time-consuming for pros, but we've found that they're still getting leads (and being charged) for jobs they don't want — and missing out on ones they do. Additionally, many pros never return to their settings to appropriately adjust the types of jobs they get.
The solution...
Our team's core focus is to rebuild the way pros build their job preferences. During H2, we developed an alpha experience that uses AI to build out and iterate on their job preferences. We'll collect feedback from their on-platform activity and use an LLM to generate new, dynamic job preferences.
For the alpha, I was responsible for creating the prompt that generates the managed memories (aka: job preferences) and building the output evaluation prompt. I also co-designed the UX experience with the product designer and wrote all the UX copy.
The results....
The pros enrolled in the alpha have had mixed experiences. It's likely this experience will work better for pros in more straightforward categories (like house cleaning). We're conducting rolling research and have started implementing changes for the beta experience, which is set to launch in December.