Tax season at a small CPA firm is controlled chaos. You know this if you've lived it. January through April, every client needs something yesterday, and your inbox is the funnel that everything pours through.
This is the story of a 3-person Phoenix CPA firm — let's call them Desert Ridge Accounting — and what happened when they stopped trying to keep up with email manually and let AI take the first pass instead.
Sarah runs the firm. She's the lead CPA. She has one staff accountant and one admin. During tax season, they handle about 180 active client files. And every single one of those clients has questions.
Here's what a typical day in Sarah's inbox looked like before:
Total: 130-160 emails per day during tax season. Off-season, it drops to about 60-80. Still a lot for three people.
Sarah was spending 2-2.5 hours per day just on email. Her staff accountant spent another hour. That's 3+ hours of combined labor — every day — on typing responses that followed the same patterns over and over.
"I'd sit down at 7am to start on returns, and by the time I'd cleared my inbox it was 9:30. My actual work didn't start until mid-morning, and then the emails would pile up again by lunch." — Sarah, managing partner [typical client quote]
Sarah isn't a technophobe. She tried the usual stuff:
Nothing stuck. The core problem was that email required both speed and context — you need to respond quickly, but the response has to be accurate and specific to that client's situation.
Sarah connected AssistantAI to her firm's Gmail accounts in February — right before the tax season crunch hit. Setup took about two days. During that time, the AI analyzed her sent emails to learn her writing style, her common response patterns, and the terminology she uses.
Here's what the new daily workflow looks like:
Sarah opens her phone and reads a one-page summary of everything that came in overnight. Emails are sorted into four buckets: urgent (needs her personally), routine (AI has drafts ready), informational (FYI only), and junk (already filed away). She scans it in 3 minutes.
She opens the approval queue. There are 18 draft responses waiting. Client document confirmations, status updates, scheduling replies. She reads each draft — most are two to four sentences, exactly how she'd write them. She approves 15, edits 2, and flags 1 for personal follow-up. Time: 12 minutes.
As new emails come in, the AI drafts responses in real-time. Sarah's phone buzzes only for the ones flagged as urgent or requiring her judgment. The routine ones queue up for batch approval. She does a second approval pass at lunch (8 minutes) and a final one at 4pm (6 minutes).
Every email has been responded to. Average response time to clients: 47 minutes, down from 4.5 hours.
Get a personalized AI briefing tomorrow morning. No card required.
Get My Free Briefing →| Metric | Before | After (30 days) | Change |
|---|---|---|---|
| Sarah's daily email time | 2.5 hours | 30 minutes | -80% |
| Staff accountant email time | 1 hour | 15 minutes | -75% |
| Average client response time | 4.5 hours | 47 minutes | -83% |
| Emails requiring manual drafting | 100% | ~15% | -85% |
| Client complaints about communication | 3-4/month | 0 | -100% |
| New clients taken on (monthly) | 4-5 | 7-8 | +60% |
The communication complaints one matters more than people think. In accounting, clients leave because they feel ignored — even if the work is excellent. Responding in under an hour, consistently, changes the entire client relationship.
Let's be honest about the numbers. Sarah bills at $200/hour. Here's the straightforward math:
Now, realistically, not all of Sarah's recovered time converts directly to billable work. Some of it goes to business development, some to lunch, some to just breathing. But even if only half converts to billable hours, that's still $4,700/month in recovered revenue against a $199 cost.
The indirect value is harder to quantify but arguably bigger. Faster response times mean happier clients. Happier clients mean higher retention. Higher retention means less time and money spent finding new ones. And the capacity to take on 2-3 more clients per month without working longer hours — that compounds over a year.
I asked Sarah what she didn't expect. Three things stood out:
1. The AI got better fast. The first week, she was editing about 40% of drafts. By week three, it was down to 10-15%. The AI learned her preferences — how she addresses different clients, when she uses formal vs. casual language, which questions she always asks in tax consultations.
2. Clients didn't notice. Not a single client commented on a change in communication style. The drafts sounded like Sarah because they were trained on Sarah's actual emails. A few clients actually complimented the faster response times.
3. The morning briefing changed everything. Before, Sarah started every day with inbox anxiety — scrolling through dozens of unread messages trying to figure out what was urgent. Now she starts with a clean summary. She knows exactly what needs her attention and what's already handled. She said it reduced her stress more than the time savings did.
"The briefing is the thing I'd miss most if I had to give it up. Starting the day knowing exactly what's on my plate instead of drowning in unread emails — I can't go back to the old way." — Sarah [typical client quote]
Honestly, it depends. If you're a solo practitioner handling 30 emails a day and you enjoy the personal touch of writing every response yourself, this probably isn't for you.
But if you're running a small firm, handling 80+ emails daily, and you feel like email is stealing time from the work that actually generates revenue — the math is hard to argue with.
The firms that get the most out of this tend to share a few traits:
If that sounds like your practice, get a free briefing on your actual inbox and see what the AI would do with your emails. No card required, takes 60 seconds to connect.
You can also check out our page specifically for CPA firms or browse more case studies to see how other professionals are using this.