The CPA firm that got its life back.

Maria runs a four-person boutique CPA practice near the Domain in Austin, primarily individual and small business tax. Roughly 240 active clients. Strong retention. Structurally burned out. Tax season meant 11-hour days, six days a week, for ten weeks. Staff overtime was a line item. New client intake was frozen January through April because there was no capacity to take anyone on.

Before any tool was discussed, the engagement opened with a problem audit. Three specific bottlenecks surfaced, each measurable.

The three bottlenecks

Document chase. Maria's team was sending an average of 3.4 follow-up emails per client per season chasing missing W-2s, 1099s, mortgage interest statements, and charitable receipts. Each email took 4 to 8 minutes to compose and track. Across 240 clients, that was a minimum of 70 hours per season in document-chase labor.

Briefing packets. Each pre-meeting packet was assembled by hand from the prior-year return, flagged changes, and open items: 45 minutes per client, for roughly 180 clients with meetings. 135 hours of prep work, every season.

Post-meeting communication. Individual summary emails after each meeting, confirming discussion, outstanding documents, and next steps, written one at a time, 10 to 15 minutes each.

The build

Three connected automations, deployed in the off-season (August through October), fully live by January.

AI document intake sequence. Clients received a structured intake form in December. The AI tracked submissions, identified missing document categories against a per-client-type checklist, and sent automated reminders every 72 hours until the file was complete. Staff watched a live dashboard. Zero manual follow-up emails.

AI briefing packet generator. A summarization tool pulled prior-year data, generated a two-page briefing per client, and flagged three change categories: income changes over 15 percent, new deductions not in prior years, and life events in the client file. Review time dropped from 45 minutes to under 10. Staff reviewed and approved; AI drafted.

Batch communication queue. Post-meeting summaries drafted by AI from a structured template plus notes entered during the meeting. Maria reviewed and approved a batch of 8 to 12 summaries each morning in under 20 minutes, instead of composing them individually throughout the day.

The outcome

First full tax season post-build: average daily hours dropped from 11 to 6.5. Staff overtime eliminated. Client response time on missing documents dropped from 4.2 days average to 1.1 days. Twelve new clients onboarded during tax season, the first time in four years.

What this engagement is, and is not

The Maria engagement is the worked example for the AI for accountants vertical guide and for the QuickBooks AI integration guide. It is not an argument that document summarization is the wedge in every CPA practice; in this case the wedge was the document chase, and summarization came second. The pattern that generalized across other accounting clients is simpler: identify the highest volume of repeated, low-judgment work, and let the senior people stop doing it.