Most firms don't start their AI journey with a fully-baked strategy. They have one task that's eating hours every week, and a hunch that something could fix it. Where AI actually fits in a practice, and which parts are worth the effort, is something firms tell us they're mostly figuring out alone.
This summer, we asked accounting and bookkeeping firms to show us what they'd actually gotten working. Not pilots or what they planned to build, but what their teams touched every day.
Firms learn faster from each other than from anyone selling them software. A firm that has already made AI work on client onboarding can save another firm six months of guessing. The accountants who solve these problems first also absorb the cost of every experiment that fails. So we helped cover that cost: $30,000 across six categories, $5,000 to each winning firm to invest in whatever they build next.
The entries answered a question firms keep asking us: where does AI genuinely help? The winners didn't automate their accountants out of the work. They took the hours that went to gathering, formatting, and re-keying, and reallocated that time to advising clients. Every one of them kept a professional signing off on the result.
The firms we learn the most from aren't waiting for the profession to settle. They're building the practice they want to run, and sharing what they learn along the way. Six winners, one per category, listed alphabetically below.
Fine Point Consulting taught its accountants to think with AI before automating anything
Community Building: firms helping other firms adopt AI.
Fine Point Consulting provides outsourced accounting, HR, tax, and fractional CFO services to growing organizations. President Chelsea Stanton's team made a firm-wide decision early, and the decision shaped everything after.
"We weren't going to start by trying to automate accountants. We were going to teach accountants how to think with AI."
The proof came from a client invoicing process that had been consuming close to 20 hours every month. Using Claude, the team designed a Power Automate workflow and an Office Script built entirely inside Microsoft's tools they already owned. Monthly processing time dropped from roughly 20 hours to under 3.
Fine Point builds their wins one client and one lesson at a time, then spreads what it learns across the firm and out into the profession.
Home Time Business Services turned a Claude prompt into a working application
Innovation Pioneer: experimental, forward-looking applications that push the boundaries of AI.
Kimberly Erick's firm provides bookkeeping, payroll, tax preparation, tax planning, and CFO advisory to small business owners. Over the past year, she brought AI into nearly every function of the practice, from standard operating procedures to advisory service development.
The most striking piece is an application she built herself, starting from a Claude skill. It reads redacted PDF bank statements and pulls transactions into a spreadsheet. It classifies each transaction against that client's business profile, accounting for industry, vehicles, home office, and known vendors. It flags what needs review, identifies likely personal expenses, and checks totals against source documents before the file returns to her team.
A single prompt became a custom application. Kimberly is clear about why the result matters. "The result isn't replacing professional judgment. It's reducing hours of manual work so we can focus on advising clients where we add the most value."
NeatBooks closes the books for 50 client locations while nobody is working
Industry Transformation: rethinking how accounting work gets done.
Olga Hurtado, EA, MBA, founded NeatBooks in Doral, Florida, to serve franchise businesses and multi-unit operators. Her team built what they call the Nightly Close.
Every night at 5:30 PM, the books close for all 50-plus client locations. Vendor bills file themselves and reconcile to the penny. Source invoices attach automatically, so the records stay audit-ready. Anything that needs human judgment gets flagged and waits.
The team arrives each morning to numbers that are already fresh and verified. An Enrolled Agent reviews and signs off. The system learns from every close.
Fifty-plus locations, and NeatBooks added no headcount to handle them. Olga built the Nightly Close as a five-step framework other firms can copy, and copying is exactly the point. "Any firm can work this way."
PMB Accounting gives small business clients the reporting that big companies get
Client Experience: AI that makes the business owner's life tangibly better.
Melissa Beck, EA, and Phil Beck run a husband-and-wife-led firm built for service businesses, contractors, and the trades. Their premise is direct.
"Small business owners deserve the same financial insight that big companies get from their CFO teams."
Using AI, PMB built a reporting system that turns each month's numbers into a branded Financial Snapshot. Clients get KPI cards for revenue, net income, gross margin, and cash on hand, each with a year-over-year comparison. Plain-English labels mark every metric as GOOD or WATCH. A revenue trend chart and a full expense breakdown complete the snapshot.
The PMB team reviews every snapshot before a client sees it. The firm stopped formatting reports and started advising. "Our people handle the interpretation, the advice, and the relationship."
Southern Payroll & Bookkeeping built 13 AI agents on top of its Gusto data
Growth & Scale: AI that helps firms take on more clients without adding headcount.
Southern Payroll & Bookkeeping serves small and mid-sized businesses in all 50 states from Chattanooga, Tennessee. CEO George Wilson's team measured how its accountants spent their days and found that roughly 80% of their time was spent gathering data and only 20% advising clients.
AI Enablement Lead Johanna Prila led the build of an intelligence layer sitting directly on the firm's Gusto payroll and bookkeeping data. Over the past year, the team built 13 specialized agents, each focused on a specific bottleneck.
The Labor Insights Agent took a 15-hour manual process down to 3 minutes. Business valuation went from 6 or 7 hours to under an hour. Bookkeeping overage review went from more than 2 hours to under 10 minutes. "Instead of producing reports, we produce decisions."
Upkeeping scrapped a training video three times, then built a reusable skill instead
Operational Efficiency: AI that makes the firm itself run better.
Erin Pohan founded Upkeeping on one idea: accounting should never be the reason a business fails.
She recorded a 45-minute training video. She scrapped it. She tried again, twice. On her third failure, she stopped trying to record and changed her approach completely.
Erin gave Claude the raw material instead. The messy POS reports, a 31-line journal entry example, and the transcript of the recording that had just failed. Once the process was right, she asked Claude to turn it into a reusable skill.
That task now takes 5 minutes a month. She applied the same method to the client communication tracker, transaction allocations, and recurring journal entries. Hours come back every month, and they compound.
"I'm not building for this month. I'm building for future me. And future me is thrilled."
What these six firms have in common
None of these firms started with the technology. Each one started with a specific task that was costing real hours, then found the tool that fit the task.
Every winner kept a professional in the loop. Olga's Enrolled Agent signs off each morning. PMB reviews every client snapshot. Kimberly's app hands work back to her team. Firms tell us human oversight is the part they worry about most, and the winners answered the worry the same way: AI handles the gathering, and your people handle the judgment.
The results also compound. Erin's five-minute process became a method she reuses across the firm. Olga's Nightly Close gets sharper with every cycle. Johanna's first agent became 13.
Your firm does not need 13 agents to start. Your firm needs one task that is costing you hours.
