40 Hours of Work Done in 5 Minutes: My Takeaways from Black Ore’s AI Tax Summit
This post is sponsored by Black Ore, which also covered my travel to the AI Tax Summit. Check them out at blackore.ai.
Black Ore claims its Tax Autopilot can do 40 hours of tax work in five minutes.
That claim set the tone at Nasdaq Tower in Times Square, where the tax profession packed into the first-ever Black Ore AI Tax Summit to answer the question: Is the future of tax autonomous?
It was standing room only. Tax leaders from EY, PwC, KPMG, BDO, Deloitte, Withum, EisnerAmper, Sax, and Elevate filled the room. Kelly Phillips Erb from Forbes served as MC. I was there to moderate a panel on the AI-native firm.
I filmed a quick recap on-site, and you can watch it here:
But three minutes wasn't enough. Here's what didn't fit in the video, and why I left New York convinced the profession isn't debating whether AI transforms tax anymore. We're only debating how fast.
The talent shortage
Eyal Shinar, Black Ore's CEO, opened the day with the numbers behind the talent shortage.
The profession is short about 125 million hours of tax work per year today. By 2030, the shortfall grows to 600 million hours. Somewhere between 300,000 and 500,000 professionals have left the field. The number of students studying accounting has dropped 60%. Meanwhile, the industry spends $150 billion a year on tax prep, and almost all of it is labor.
We don't have enough people to do the work. The math only works with AI.
What stuck with me from Eyal's keynote was his framing of why generic AI tools haven't solved this already: "Tax is an ontology problem, not a language problem." Bolting an LLM onto tax software doesn't work. It hallucinates. And much of the software it would bolt onto still runs on COBOL, patched for decades.
So Black Ore built something different. The five-minute claim I opened this post with comes from their flagship example: a complex high-net-worth return with dozens of K-1s, work that normally takes a senior associate about 40 hours, prepped by their Tax Autopilot. Their published numbers are 99%+ accuracy, 98%+ autonomy, 98%+ time savings, 80%+ cost savings, and 2x faster reviews. They say returns prepared this way have already been audited by the IRS.
Where the ROI is showing up
The most convincing evidence at the summit wasn't a demo. It was financial results.
Jeff Wong, the former Global CIO at EY, said the firm generated $6.50 of revenue for every $1 invested in AI last year. That wasn't cost savings. It was net new revenue. But he cautioned that if you don't change your pricing or introduce new services, the efficiency gain will just evaporate.
Ryan Stevens, Director of Applied Science at Ramp, brought data from the 50,000+ U.S. businesses whose spend is processed by Ramp. Companies with high-intensity AI spend are growing revenue 6 to 7 times faster than their peers. And they're not seeing widespread layoffs at AI adopters.
Stevens also offered the sharpest disruption thesis of the day. Job disruption won't come from AI inside the big firms. It will come from 10-person AI-native firms taking work from 3,000-person traditional ones.
The leverage gap in firms
David Frigeri, Chief AI Officer at EisnerAmper, laid out four tiers of AI leverage:
No AI: Firms not using AI at all operate at 1x leverage.
Tools only: Firms that buy tools without changing their processes get 1.2 to 1.5x.
AI-first: Firms that redesign their processes around AI get 5 to 10x.
AI-native: Firms built from scratch around AI get 20 to 50x.
He says that currently only 7% of firms are AI-first. Only 1% are AI-native. Everyone else is buying tools, changing nothing about how they work, and getting a 20% bump.
That gap is the story of the next five years. The difference between a 1.5x firm and a 20x firm isn't software spend. It's whether you're willing to redesign the work itself.
The Big Four tax technology panel, moderated by K2's Randy Johnston and Brian Tankersley, backed that up with insights from inside the largest firms. Asked what they'd fix first with a magic wand, every panelist gave the same answer: data. Tax professionals spend 40% to 70% of their time wrangling it. And the biggest obstacle to AI adoption inside a firm isn't the staff. It's the partners.
From 15 years to 15 months
My panel was "The AI-Native Firm: Unifying Tax, Advisory, and Assurance," with Jim Bourke of Withum, Becky Munson of EisnerAmper, Sean McLean of Elevate CPA, Ryan Hittner of Deloitte, and Geni Whitehouse of the Information Technology Alliance.
What Becky Munson said stuck with me: "Cloud took 15 years. AI is happening in 15 months." Jim, who has steered Withum through paper, on-prem, cloud, and now AI over the past four decades, agreed that this technology shift feels different. He explained why big firms move slowly: "The bigger the ship, the harder to pivot."
Sean McLean put numbers on that asymmetry. A small firm can implement a tool like Black Ore in 2 to 4 weeks. A big firm needs 18 to 24 months.
He also told my favorite story of the day. During the recent tax season, firms put AI in front of their preparers — and the human preparers still went back and double-checked everything. The firms paid for the AI tokens AND the labor. Costs doubled. As Sean put it: "CPAs love to reconcile until the cows come home." Adoption isn't a technology problem. It's a trust problem.
Ryan Hittner made that concrete from the audit side: there are no AI auditing standards yet. Auditors are operating in a gray space, and trust will be built the way Tesla's autopilot built trust — qualitatively, over time.
And Geni Whitehouse made an important point: AI is finally going to unlock the advisor role this profession has been talking about for 20 years. Compliance work always absorbed the bandwidth. Now the constraint shifts to human skills — storytelling, asking the right questions — and those have to catch up.
The uncomfortable part
Allan Koltin closed the day. His headline prediction was blunt: "By May 20, 2030 — four years from today — accountants will no longer be preparing financial statements or tax returns."
He predicted that 80% of the current work in accounting will drop to zero in value. But he also said the remaining 20% will grow 10,000 times in value. And the Big Four are already acting like they believe it. They cut on-campus accounting hiring by 50% in 2020, then cut it another 50%. They're hiring STEM grads and training them up instead. Partners no longer have tenure-like job security. Layoffs are hitting every level as firms restructure for an AI-driven future.
Is Koltin right about the date? I don't know. 2030 is an aggressive timeline. But notice that his prediction is directionally identical to everything else I heard that day, from the keynote math to Ramp's growth data to Frigeri's leverage tiers.
But Koltin isn’t pessimistic about the profession. He also believes that "there has never been a better time for kids to go into public accounting."
The 20% that remains is the interesting work: the judgment, the relationships, and the strategy. That's the work everybody actually wanted to be doing when they signed up for this profession.
My takeaway
If you think you have time, you don't.
Becky's prediction of 15 years versus 15 months may be bold, but it points in the right direction. The cloud gave firms decades to adapt, and some still haven't. AI is moving faster. We won’t have that luxury this time.
The good news is the destination. This profession has talked for a generation about moving from compliance to advisory, while compliance ate every available hour. The firms that redesign their work around AI get to keep the part of the job people actually like.
The profession isn't debating if anymore. It's debating how fast.
