Pros today miss out on a lot of money due to invoices not getting paid. They go into collections, which is a painful and time consuming process. Pros need to manually send reminders to their customers, which often results in a poor customer relationship. There is a fine line between following up on payments and pestering your customers too frequently. This seemed like a perfect task for an agent to handle, since it can learn from success rates with factors like time to pay, delivery methods, collection prevention and more.
On top of that, there's a trust problem: getting a Pro to default to letting an agent fully act and chase up invoices on their behalf, with no manual review, means handing over something sensitive to a system they haven't had the chance to build confidence in yet.
I led 3 individual interviews with our Pros to better understand how they go about collections. It was clear that an automated solution is highly desired. Housecall Pro has a lot of features buried in settings, and one of them was automations for overdue invoices. Granted, the setup experience was not great, but the functionality existed.
I looked at the end-to-end user journey of starting a job, to sending an invoice and needing to collect. Many opportunities came out of the user journey and as a team we decided to focus on invoice collections as a starting point. I wanted to surface an invoice reminder flow on the invoices page, a page that sees high traffic and is most relevant to collections. I designed my prototypes in Claude and shared with the team for feedback often. The direction was loved by stakeholders, and I decided to build it out even further.
I wanted to make the experience feel valuable and unique, compared to sending out collection reminders manually. I added a tone-of-voice feature that would generate a reminder message based on the amount of days an invoice was overdue. The longer past due an invoice, the stricter the tone of voice would become. AI would automate these messages based on due date, and all a Pro has to do is click send.
We made following up on unpaid invoices much easier, and were able to build enough trust for Pros to automate the feature entirely.
The goal was that Pros would eventually trust the agent so much that they'd click a button to automate the entire invoice follow-up flow, and never have to think about collections again.
We got there by rolling out trust in stages: Pros started by reviewing every AI-drafted reminder before it went out, and once they saw the tone-of-voice logic consistently making the right call, more of them turned on full automation. Reminders that used to feel like an awkward, manual chore now go out on their own, tuned to how overdue an invoice actually is.