AI reporting added to a system already in use, in one week

The problem

The practice was already running the time logging app I built, reporting included. But questions kept coming up that the reports weren't built to answer. Each one meant the COO asking me to add a new button or filter, then waiting for it. That was slow for them and not a good use of time for either of us.

What I built

An AI assistant built into the app, in two parts.

First, a chat inside the app. The COO asks a question in plain English and gets an answer from the practice's live data. I set it up with the practice's own terms and the way their numbers are calculated, so it answers like someone who knows the business. It can also produce reports, presentations, and other files on request. In practice, it works like an analyst who is always available.

Second, I upgraded the reporting page. It now has a drag-and-drop report builder, and reports can be saved and run again later. The assistant can use that builder too, so instead of building a report by hand, the COO describes what they need and the assistant builds and saves it. Several reports can then be combined into one PowerPoint or PDF presentation.

The practice's data stays in its own AWS account and is never used to train AI models.

Results

Reporting went from a fixed page that needed me for every change to something the COO can simply ask. New questions get answered right away instead of becoming a request, useful reports are saved for reuse, and presentations come straight out of the app.