AI for the Tax Pro, or The Cart Before the Horse

Hello and welcome once again to The Friday Edition. I’m an international tax attorney who writes this for other international tax pros – accountants, enrolled agents, and attorneys. Thanks for subscribing.

Upcoming

• Second Annual International Tax Summit. London, June 26, 2026. A full day of international tax topics chosen for practitioners who serve individual and family clients in cross-border tax advisory and compliance. CPE and CE credits available. Members of the International Tax Pros community sign up for the discounted ticket inside the community. Nonmembers are welcome; buy your ticket here.
• (Tentative) The Business of Practice. London, June 27, 2026. A half-day session. Put on your practice management hat, your business owner hat. How do we make more money with less stress? How do we deploy AI and other tools in our businesses? See the discussion at the bottom of the newsletter. Indicate your interest (no commitment!) here.
• There is no International Tax Lunch this month in observance of the vacation everyone needs after the tax deadline. If you’d like to save the date – the next session is May 29th, 9AM PT. Debra will be giving an expatriation flavored presentation.

You Wear Two Hats

Usually I talk about technical tax topics. This week’s newsletter focuses on AI and the international tax pro’s role as a business owner/operator.

Tax pros wear two hats: technician and operator.

• We are technicians–we know how to solve tax problems. We produce the product that our clients pay us for.
• And we are also operators. We run a business–sometimes as small as one person, sometimes larger. We do the marketing, the sales, the HR functions, and everything else. Even tax pros who work for a firm are business operators, but that’s a conversation for another day.

AI is The Current Thing, and it affects us as technicians. LLMs are becoming semi-reliable substitutes for books and online research services to answer technical tax problems. We can do more, faster. With a fair degree of confidence in the answer–if we check every detail.

I would argue, however, that the real impact of AI is for what you do as an operator. Tax questions need to be answered precisely. Business decisions usually just need to be directionally correct and good enough. An LLM will deliver the median solution–and that’s all you usually need.

So let’s talk about AI for the international tax pro as an operator: running a business.

AI is Useful When You Know What You Want

AI is a useful tool only when you know who you are and what you want. By this I mean your business. Your firm exists to give you the kind of life you want to live, and you get to choose what life you will live. If you want to spend time with your kids but you have a business that demands all of your time, there’s a problem. Automating that business exacerbates the problem.

You need to know your Main Quest. If you are adding employees, providing new services, adding overhead, is that your Main Quest in life? It’s perfectly fine to build a juggernaut professional services firm, but if that’s not your heart’s desire, you are making your life worse, not better. Will more money (you hope) compensate for more stress, more risk, less time with your children?

Values –> Policy –> Process –> Tech

I have been self-employed since March, 1991. Since then, I have fumbled blindly in the general direction of optimizing for Life rather than money. I started (like everyone) with “Any dollar of revenue is a good dollar” and got smarter through pain. Much of the self-inflicted pain came from searching for the latest Shiny Thing, which invariably meant software.

A podcast (YouTube) recently crystallized the process for me. And it told me why AI (and indeed any other software) comes last. I recommend listening to the entire podcast.

The hosts explained the folly of searching for business success by asking “What software do you use?” That’s the last question you should be asking.

Here’s the sequence that they suggest:

• Values. What do I want my business and my life to feel like?
• Policy. What is my standing policy when x event arises? (I.e., pre-decide how you will act when something happens, so that your response will be consistent with how you want your business and life to feel like).
• Process. What are the steps you are going to take to execute that decision you pre-decided? (E.g., standard operating procedures).
• Tech. What tools support the process?

Example: Intake and Onboarding

At HodgenLaw PC we have successfully thought through the Values and Policies for work we want to do and clients we want to work for. Over more than a year, we worked out the Process for what happens from the moment of first contact by a prospect until we take on the job or not. And the whole pipeline is managed by Tech, which in our case is Nutshell. And by “we” I mean mostly Maeghan Gershey.

If you don’t know what type of work you want to do, any prospective client is enticing. If you don’t know what types of people you want to surround yourself with, you won’t be able to filter in the right people and filter out the wrong people.

If we had started with “I want to design an intake process for prospective clients” typed into an LLM, we would have been handed a Process, and possible we would have been told the right Tech to use (maybe a vibe-coded app, maybe a commercial SaaS).

But Claude would not and cannot decide what we collectively as a firm want our business and our lives to feel like. And Claude cannot tell us how to pre-decide our response when a prospect asks us to vary our standard intake process.

AI cannot decide Values or Policy. AI can only create Process and Tech. And if you start with Process and Tech, you are automating a process to create results that may make you unhappy.

How To Deploy AI as an Operator?

AI is a thing. And it is possibly a useful thing. Don’t ignore AI. Know what you want and use AI to help you get it.

Field Reports, Not Opinion

It’s hard to tell the AI bullshit from the AI jellybeans. Cedric Chin has an article that tells you how he makes sense of AI. In brief: look for field reports. Ignore opinions, noise, hype, etc. and instead look for detailed reports from actual people doing actual things with AI–successfully or unsuccessfully.

When reading field reports, ask yourself these questions:

1. What new outcomes are implied here?
2. What actions may I take?
3. What are the relative value of those outcomes to me?
4. What cause & effects?

Outcome Orientation

Cedric also has an idea he calls Outcome Orientation. Anytime you catch yourself doing something, ask yourself “What is the outcome I am trying to achieve here?” If you know your Values and you have thought about your Policies, you will know what your Main Quest is. Then you simply decide whether the action you’re engaged in at the moment is in service of the Main Quest.

I successfully resisted the siren call of OpenClaw earlier this year just by using Cedric’s methods.

Deploy AI in direct service of your Main Quest. I am inside Claude Code (on the command line) and Obsidian every day, but only for tasks that accelerate my Main Quest. And when it does, I am willing to spend 5x the amount of time working with AI to accomplish my Main Quest.

Time spent on Side Quests is wasted time.

Share and Learn

Cedric suggests belonging to a community where AI field reports are shared. This solves the problem of “honest field reports are rare and almost everything on the internet is hype.” I belong to two such communities: Common Cog (Cedric’s community) and Dynamite Circle. It’s simultaneously awe-inspiring to see what others are doing and reassuring to see that we’re all fumbling around right now.

In the International Tax Pros community, we are starting to see this as well. Travis Call, CPA gave a presentation last month on how he uses multiple LLMs to do tax research. Christine Dahl is showing us how she automates her practice one piece at a time.

Soft Sell: June 27, 2026 – The Business Summit

A soft sell here.

The London International Tax Summit is on Friday, June 26, 2026. That’s a full day of technical tax presentations.

I’m thinking of organizing a half-day session organized in London on Saturday, June 27, 2026. Location in London TBD.

The objective of this half-day is to get together in an informal and loosely-structured small group to talk about operations. How do we run our firms? How do we make more money with less stress? How do we deploy AI in our firms? This will be very much a “bring your laptop” event.

The Saturday session will be free for members of the International Tax Pros community, and will cost money for nonmembers. I don’t know how much yet, because that depends on the venue and how much money they want for the size of room that we will need.

If you’re interested in attending the Saturday session, please let me know by signing up in this little landing page. Just name/email address.

No commitment. I want to know how much interest there is – so we can get the right sized meeting room.

You don’t have to attend the Friday International Tax Summit in order to attend the Saturday event where we focus on operations. But I hope you do attend both!