I spent this week at AICPA Engage in Las Vegas — my first time, and as a CAS practitioner who isn't an American CPA, I wasn't sure what I'd find. I half expected an older crowd working through technical updates. Engage is an amalgamation of what used to be many smaller conferences, so the technical tracks are certainly there — but what I found was an energizing blend of people and interests, and a profession finally saying the quiet part out loud.
The main event opened with a town hall from AICPA leadership. Right from the start, they acknowledged that adapting to technology and AI is the top issue for CPA firms of all sizes, with staffing a close second. Sit with that. The profession's own leadership, on its biggest stage, put AI at the top of the list — ahead of everything else the profession usually argues about.
The keynote slot tells you everything
The first keynote went to Clara Shih, on the organizational reset underway as roles get redefined into managers of agents — employees who orchestrate a set of specialized AIs toward a goal rather than doing every step themselves. Her talk had almost nothing to do with accounting and everything to do with the pace of change from the consumer layer up through the organization. If you've been reading this channel, none of it was revelatory. But that's not the point. The point is that the very first keynote of the profession's flagship conference was about AI reshaping every business — including yours. Conference programming is a confession of priorities.
What the vendor hall said about open walls
On Monday I said I'd be asking vendors the open-walls question to their faces. Here's the short version: nobody on the floor has open walls yet.
The hall held a hundred-plus vendors, and nearly all of them had AI somewhere in the pitch. On the ledger side, Puzzle and Digits stole the show — accounting ledgers rebuilt from scratch around AI, with AI-based transaction coding and built-in agents for accountants. Neither is ready for my firm: multi-currency and sales tax handling aren't there for the Canadian market, and both say openly they're not ready for Canada. For US firms, though, these products could now serve a good number of clients. The incumbents didn't look great by comparison — Xero's booth was mostly about its Melio acquisition in payments, and the QuickBooks staff couldn't answer AI questions. Not a good look for either platform.
But here's the integration verdict, and it's the part that matters. Both Puzzle and Digits claim APIs, but they're built to work inside their own ledger, not across ledgers — their integrations pull data in, they don't let an outside system interrogate or instruct them. Combinely, which demonstrated an intriguing email-driven trigger system for bookkeeping automations, works on the same basis. And Artifact — whose two products sit above your existing QuickBooks, Xero, Sage, or NetSuite ledger to orchestrate month-end close — shows no mention of an API or an MCP anywhere. (MCP is the emerging standard that lets an outside AI, like Claude, plug into a system and work it directly rather than just receiving its exports.)
These are salespeople answers, and I'll reserve final judgment until I've done real testing. But the pattern was consistent: data flows in, control doesn't flow across. Every tool is building agents inside its own walls — and what your practice actually needs is tools that work together.
2030 is too late
Jody Padar — The Radical CPA — ran a session called "The Firm of 2030 is not Today Plus AI." Her model of the future firm is a barbell with a brain: AI handling lean execution on one end, humans leading expanded advisory on the other, connected by a central intelligence layer focused on strategy and data storytelling. The AI-ready firm gets built on four layers — secure client collaboration, workflow automation, integrated systems, and AI-assisted advisory. And the value proposition shifts from selling outputs — filings, statements, compliance — to delivering outcomes: decision clarity, confidence, and risk reduction.
I agree with nearly all of it. My problem is the date. I think we have one more busy season, maybe two, before external pressure makes the old way of doing business unworkable — and firms that haven't adjusted by then won't fade gradually, they'll become irrelevant to the marketplace while compliance work reprices around them. I hope 2030 is the real deadline. But if 2030 is where you set your sights, you'll arrive late. The target is 2028.
I wasn't alone on that timeline. Mark Stout of Revere Consulting laid out a 2028 vision of AI-augmented professionals — and urged firms to act within the next 12 months to avoid falling behind. That's the timing I believe is most critical.
Buy versus build just flipped
Nicole Davis and Chris Hervochon ran a session called "Not Another Tech Stack" with the most practical strategic advice of the week. For complex core systems — general ledger, tax software, payroll — never build your own. For nearly everything else — advisory, analytics, reporting, proposals, marketing, client portal, practice management, workflow automation — building has become accessible enough that they now advocate building over buying. Look at what buying actually gets you: most firms use maybe 40% of a purchased app's features while wishing for a list the vendor hasn't built. Build your own and you get 100% of what you want and none of what you don't — if you're willing to own the maintenance and security that come with it. I spoke with plenty of practitioners already experimenting down this road.
It connects to the warning I heard repeated in session after session: buying a tool is not a strategy. Every vendor pitch on that floor was fantastic, and every demo looked excellent. But until you can answer what problem you're solving and where the tool sits in your client service delivery, you aren't ready to buy.
The human counterweight
The second keynote put Ryan Reynolds — Canadian, I'll note for the record — alongside Richard Galanti, Costco's long-time former CFO. (The connection involves one of Ryan's films and Costco hot dogs. I promise it made sense in the room.) The conversation was about leadership, barely touched AI, and may have been the most important session of the week for an AI-focused attendee.
Both kept returning to constraints, and how limits breed creativity and focus. AI removes constraints — that's mostly the point of it. But the tool that breaks you through a constraint can quietly take over the thinking the constraint used to force. Both also talked about culture and the customer: Costco's core asset is customer trust, and Reynolds is rigorous about his audience's experience across every venture he touches.
That's the counterweight to the managers-of-agents future. If your people are going to orchestrate AI, their culture, growth, and engagement still have to be led — or your best staff become tools of the machine, and leave. The same holds for clients. As delivery becomes more software-driven, your clients never want to be another bit of data feeding the system — they want to feel valued and heard, and most want the relationship above the results.
The advisory sessions backed this up. CPA.com — whose education program impressed me all week — is now talking about the zero-day close as an achievable target rather than a utopian dream, and a strong session on advisory at scale pushed proactive over reactive, repeatable processes over generic ones, and technology-enabled personalization. One speaker put it best: "Cadence without warmth feels robotic, but warmth without cadence feels like a loss." Automation buys you the cadence. The warmth is still your job.
What to do with all this
Here's my read after four days on the floor. The profession has agreed on the destination — AI-augmented firms, advisory-led, built on integrated systems. The disagreement is the timeline, and I'm telling you the deadline is 2028, not 2030. The vendors will sell you pieces of the destination, but none of them yet will sell you the connected system — and a tool without a strategy is just another subscription.
So do the strategy work first. Decide what problems you're solving and what your client delivery should look like, and only then pick the tools. That's exactly what our AI Practice Transformation program is built for — the next cohort starts June 26th, and you can book a call with me at theaiaccountant.ai/transformation. And if you're at Scaling New Heights in Orlando next week, message me on LinkedIn — I'd love to meet.