Taxwell’s product designers run up to 10 AI-moderated customer conversations in a day or two, then use what they learn to shape the next iteration. It’s a way to keep customer feedback close to the work, even as the team moves quickly from one product idea to the next.

The challenge: A short season and no researchers

The company makes tax software for two very different audiences: everyday DIY tax filers using TaxAct software, and tax professionals using Drake Software. Their work runs on a calendar. Consumer season peaks from January through April 15, and then extension deadlines on September 15 and October 15 keep the pressure on.

The windows are short, and the team doesn’t have dedicated researchers. Andy Vitale, VP of Product Design, leads full-stack product designers who handle content, design, and research themselves. Spending eight hours on customer calls for every idea wasn’t an option.

Taxwell is also moving faster than ever, releasing zero-to-one products and updates at a steady clip. Research had to keep up. “I think the MCP and the AI mod have really kept up with us as we’ve been trying to release zero-to-ones and just new updates at a steady pace,” Andy says.

The solution: always-on research, built where the product designers already work

Taxwell’s product designers build mostly in Cursor, and those Cursor prototypes are what they test with. “We’re prototyping with HTML, not Figma files,” Andy says, and the team built a way to host those prototypes externally, so “we just have to give a link.”

Dscout’s MCP fits into that workflow. “It’s easy to stay in that environment to get a prototype ready to test,” Andy says. “So we just start prompting the MCP and it builds the studies for us.” 

"That level of support tied with the AI moderation really takes a lot of the stress off the product designers."

- Andy Vitale, VP of Product Design, Taxwell

The team has used the MCP for AI-moderated user testing, planning, and screeners. “The screeners have been really helpful for us,” Andy says. On the professional side, the team works with Dscout to make participants “prove they use the software.”

The team also uses Dscout’s AI-powered “Draft Your Study” feature a lot. And because Taxwell doesn’t have dedicated researchers, Dscout’s review before launch matters. “Having someone review our surveys before we send them out makes them a lot more successful,” he says. “That level of support from Dscout, tied with the AI moderation, really takes a lot of the stress off our product designers.”

Taxwell product designers stay in the environment where they’re building, and “being able to be in the environment that we’re building and go right to testing is fantastic for us.”

The team’s loop is simple: “idea, create something tangible, get internal and external feedback, and then iterate, and then let’s get this thing out the door.” Dscout sits inside every pass. “It’s a lot of little feedback loops,” Andy says, “and Dscout’s at the heart of that for us.”

Andy’s team had experimented with AI moderation with another vendor for over a year. When Dscout announced that AI-moderated studies were now available to beta users, they decided to keep all their work on one platform.

The results: Speed without a speed bump

With AI Moderator and the MCP, Taxwell gets customer feedback within a day or two and keeps conversations flowing even in the off-season, when hard-to-reach professional audiences are thin.

AI Moderator is also surfacing things people might not say to a person. As Andy puts it, “We’ve found in some cases that people are a little bit more embarrassed to share things person-to-person than they are person-to-AI moderator. So that’s helped us uncover some interesting insights.”

Just as important, the experience feels smooth for participants and for Taxwell’s team. Follow-up questions are where AI moderation can go wrong, with tools that keep probing until a participant is ready to quit. Andy’s team has seen the opposite. “With us, the follow-ups have been really, really smooth,” he says. “We’ve gotten no complaints, which is ideal.”

Dscout’s optional pre-launch review of studies, paired with AI Moderator, also takes pressure off product designers who aren’t formally trained researchers. “That level of support tied with the AI moderation really takes a lot of the stress off the designers,” Andy says.

For a team moving at a pace the company hasn’t seen before, research is an accelerant, rather than a speed bump.