The Most Popular Thing We Give Away Is Not An Agent

Anybody in the accounting profession can take one of our AI agents free. No card, nothing renews, and there are close to two hundred to choose from: https://aiaccounting.legacysbc.com/claim-agent

So I have four months of people walking up to a shelf of two hundred things, being told to take whatever they want, and picking one. Seventy three claims. Sixty eight different people. I pulled the whole list this week and went through it properly, because what people pick when nothing is being sold to them is more honest than any survey I could run.

The result was not what I expected, and it changed my mind about something.

The three most claimed things are not agents

Here is the top of the list.

The AI Accountant Mega-Toolkit. Eight people.

50 Prompts That Replace Busy Season Overtime. Seven people.

Complete AI Operating Playbook. Six people.

Twenty one of the seventy three claims, and not one of those three is a working agent. They are maps. A kit, a list, and a playbook. Add in the 90-Day Implementation Roadmap, the AI Maturity Scorecard and the ROI Calculator and it is twenty nine claims, thirty nine percent of everything ever taken, spent on figuring out the terrain rather than on a tool that does a job.

The fourth most claimed is the Bookkeeping Catch-Up and Cleanup Diagnostic, which is a diagnostic. It tells you how bad it is.

So the top four things this profession reaches for, given a completely free and completely unconstrained choice, are three maps and a diagnosis.

I want to be careful here, because that is not a criticism

My first reaction was that people were stalling. Taking the reading list instead of doing the reading.

I have decided that reaction was wrong, and I think the behaviour is exactly right.

You cannot browse two hundred things. Nobody can. When someone hands you a catalogue of two hundred agents and says pick one, the honest answer is that you do not yet know enough about the terrain to know which one matters, and the rational move is to take the map first. I would do the same. I have done the same, in every field where I was the beginner.

And there is a second thing in it. A toolkit is safe. It cannot fail on Monday, because it does not have to run on anything. A Bank Reconciliation Agent has to actually work on your actual bank feed, and if it does not, you have learned something uncomfortable about how far behind you are. The map has no such risk. So of course the map wins.

But a map is not a mile

Here is what I have to say honestly about our own free products, and it is not good salesmanship.

The Mega-Toolkit will not close a month. The prompt list will not reconcile anything. The playbook will not chase a single missing document. They are genuinely good, people ask for them more than anything else we make, and not one of them will change what lands in your inbox on Monday morning unless you go and build the thing they describe.

That is not a flaw in them. It is what a map is for. But it does mean the most popular thing we give away is the thing least likely to change your week, and I would rather say that out loud than let sixty eight people quietly discover it.

There is a number that makes this plainer. Thirty different agents have ever been claimed. The other hundred and seventy or so have never once been picked. Not because they are worse, but because when you hand somebody a catalogue, they take the catalogue.

What people took when they did take a real agent

This is the other half of the list, and it is worth your attention because it maps the pain of this profession with no marketing in the way.

Twenty four claims went to what I would call the mess. Bank Reconciliation. Month-End Close. Document Collection Chase. Chart of Accounts Cleanup. Credit Card Reconciliation. Adjusting Journal Entries. Vendor Statement Reconciliation. Expense Categorisation. AP Bill-Pay. And the Catch-Up and Cleanup Diagnostic, five times, the most claimed working tool on the entire list.

Twenty claims went to everything forward-looking put together. Forecasting, proposals, dashboards, growth, tax planning.

So the mess beats the future, and it is not close.

Sit that next to what our profession markets to itself. Five years of move up the value chain, stop being a compliance shop, become an advisor. Then look at what people reach for when it is free and nobody is watching. Chart of accounts cleanup and a document chaser.

I do not think the advisory message is wrong. I think it skips the step everybody is actually standing on. You cannot advise on top of a broken chart of accounts. You cannot forecast from books four months behind. You cannot talk margin by job when the job costing was never set up. When somebody picks cleanup over forecasting they are not being unambitious, they are being accurate about where they are standing.

Put the two findings together and you get one sentence

People want a map, and the thing they want the map to lead to is the mess.

They know where the pain is. They know they need a route to it. What almost nobody has is the middle bit, the part where the route becomes a thing that actually runs on their own files.

That middle bit cannot be downloaded. That is the honest reason it is missing from all three of our most popular free products, and it is exactly what next week is.

The free week, plainly

One hour a day, Monday to Friday, live at 3 PM CST. Every session recorded. No card, nothing renews, and no coding at any point.

Three steps across the week, in this order deliberately. Train, so you understand what these things are and where they break, with a person answering your actual question rather than a PDF. Build your operating loops, meaning the repeating work in your own practice gets turned into something that runs. Then run the transformation plan on real work.

That is the map and the mile in the same week, which is the only arrangement I have seen work.

You keep an agent for every day you show up

One AI agent per day you attend live, up to five, yours to keep.

Five agents cost ninety nine dollars to buy outright. So the plain arithmetic of next week is that we hand you ninety nine dollars of product for spending five hours with us.

And unlike the toolkit, you will have built one of them on your own work, which means you will know what to do when it breaks. That is the difference between owning a tool and owning the ability to make tools, and it is the entire reason we run the week rather than just mailing out the playbook.

Why it is free, since people always ask

Usually with a slightly narrowed eye, which is fair.

The answer is not generous. It is that demos do not work. I have done demos. People nod, they say it looks impressive, and absolutely nothing changes on their Monday. Sixty eight people have now proved the softer version of the same thing to me by downloading a map and, in most cases, stopping there.

What works is an agent running on your own file, producing your own numbers, finished. You cannot un-see that. Nobody has ever un-seen it and then decided the old way was fine.

So we are not selling the week. We are proving the thing in the only way that has ever landed, and the agents you leave with are yours whether you ever buy anything from us or not.

Bring somebody. Each person registers individually, and if you have staff, two people who bring it back into the practice is worth more than you going alone.

Registration closes tonight

September 7 to 11. Registration closes tonight at 11:59 PM Eastern. Day one is Tuesday the 8th, because Monday is Labor Day and there is no live session.

https://humansplusagents.ai/legacy-sbc-aia

If you miss tonight, the next ones are September 14 to 18 and September 28 to October 2 on the same page. But I have just spent a week looking at seventy three people who took the safe option, and I would gently point out that "the next one" is the safe option too. It is where a thing you have been meaning to get to goes to die.

You do have to register to get the replays. They only go to people who signed up. Somebody messages me after every single week asking for the recordings and I have to say no, and I hate that conversation.

If a week is genuinely more than you can commit today, then take the free agent, and take a working one rather than a map: https://aiaccounting.legacysbc.com/claim-agent

And the free practitioner community is open to anyone, no cost and no pitch: https://www.facebook.com/groups/1024923770013348

One last thing

Sixty eight people. Two hundred choices. The three most wanted things we make are a kit, a list and a playbook, and the most claimed working tool is the one that tells you how far behind you are.

Nobody in this profession is confused about the problem. What has been missing is a route to it that does not cost an hour per client and a developer.

That is next week, and it closes tonight.

AI does the typing. You make the judgment calls.

Yvonne Razo, Co-Founder, AI for Accounting

Bon's Place

Friday, September 4, 2026