Happy Thursday, everyone — Albert here, and we have a full show today covering a $24 million bet on AI delivery through IT providers and what that actually means for your wallet; two-thirds of small business owners now claiming they use AI and why that number is much more complicated than it sounds; a data breach at one of the largest CPA firms in the country and what it means if your accountant’s IT is an afterthought; and a privacy problem inside Notion AI that is a good reminder about what you should never type into a chat window.
Let’s get into it.
Hats AI Raised $24 Million to Sell AI Through IT Providers — Here Is the Catch
A company called Hats AI just raised $24 million to build an AI platform that gets delivered specifically through managed service providers like me. The pitch is that large enterprises have figured out AI, and now it is time for small businesses to catch up. One dashboard, all the major AI models, full governance over what your employees can and cannot do. Sounds tidy.
I visited their website. The marketing copy is polished. And I will give them this: the idea is not wrong. Small businesses do need a way to manage AI access, control spending, and keep employees from going rogue with whatever free tool they found on their lunch break. That problem is real.
But the pricing model buried underneath all of that is the part worth paying attention to. When a platform aggregates multiple AI models into one interface and lets you switch between them, it is almost certainly running on API-level pricing. And API pricing is the most expensive way to buy AI. You are paying per token, per call, at rates that add up fast — especially if you are doing anything beyond writing a quick email.
Compare that to just buying a subscription directly. For around $200 a month on Claude, you get an enormous pool of tokens. Same story with the other major providers. The jump from the base tier to the $200 tier typically gets you something like ten times the capacity. I have been running at that level and I have never run out — except with Grok, which eats tokens like it has a personal vendetta against my budget.
So I built something. I am not going to oversell it because I am still working through it, but the core idea is a bridge that lets me pick whichever AI I want to work with at the top level, then intelligently hand off the actual heavy lifting to whichever model does that work best and cheapest — all running on subscription pricing instead of API pricing. My agents can talk to each other, pass prompts back and forth, and I can route expensive compute tasks away from the pricey environments. If I had just stayed inside one API-priced environment for everything I currently do, I am estimating the cost would be in the thousands of dollars a month. With subscription routing, it is a couple hundred.
If that sounds like something useful for your business — managing your team’s AI access, controlling what they can do, keeping the spend sane — let me know. I think there is something real here, and I would rather build it right for clients than point you at a $24 million marketing deck.
Two-Thirds of Small Businesses Say They Use AI — What That Actually Means
A new survey says two-thirds of small business owners are already using AI. A year ago, similar surveys were showing 20 to 30 percent. So we apparently doubled adoption in twelve months.
Alert readers will immediately ask the right question: what counts as “using AI”?
If opening ChatGPT once a week to clean up an email qualifies, then sure, two-thirds sounds right. But that is not really using AI — that is dipping a toe in the shallow end and calling yourself a swimmer. The actual capability of these tools is so far beyond “make this paragraph sound better” that it almost does not deserve the same category.
I have helped a handful of people get set up recently. One is a salesperson who can now use AI to find new markets — straightforward for me to configure, transformative for him. Another needed to manage a website through Claude and just could not wrap her head around the workflow until we walked through it together. Both of them are now doing things they could not do before. But neither of them is close to using even a fraction of what these tools can actually do.
The adoption number I find more interesting is what is happening on the infrastructure side. We just launched a locally-hosted backup service for our clients — we own the server, we own the configuration, we are not routing support tickets through some black-box vendor halfway around the world. AI made that possible for a company our size. We can now support our own products, on our own terms, with our own people. That is the kind of adoption that actually changes how a business runs. Not a chatbot. Ownership.
Baker Tilly Had a Data Breach — Check Who Handles Your CPA’s IT
Baker Tilly, one of the largest advisory and CPA firms in the United States, is notifying clients that personal information was compromised in a data breach.
This one stings in a specific way. You did everything right. You hired a reputable, well-known firm. You trusted them with your financials, your tax documents, your business records. And your data still ended up exposed — not because of anything you did, but because of a gap somewhere in their security posture.
I will be honest: I know from experience that IT security is not always high on the priority list at accounting firms. The focus is on the accounting. The IT is often an afterthought, handled by whoever is cheapest or whoever has been there the longest. Baker Tilly is a big operation, and big operations can have big blind spots.
The practical question for you this week: do you know anything about how your CPA handles data security? Not their reputation as accountants — their actual IT practices. It is worth a conversation. One question is enough to start: “What do you do to protect client data?” The answer will tell you a lot.
The encouraging part is that AI-assisted security monitoring has changed what a small, focused IT team can see and respond to. Things that used to require a large security operations center are now manageable at a much smaller scale — if the team is paying attention and not stretched too thin. Baker Tilly sounds like a case where the scale got ahead of the attention.
Notion AI Exposed Chat Transcripts — One Rule Worth Repeating
Notion AI had a breach. Chat transcripts were exposed.
One rule, and it has not changed since the first chatbot went live: never type anything into an AI chat window that you would not want on a billboard. Passwords, API keys, client Social Security numbers, proprietary pricing, anything sensitive — it does not belong in a chat. Not because the AI is going to do something malicious with it, but because chat logs are stored, and stored things get breached. Or read internally by someone you did not intend. Or subpoenaed. Or just sitting there waiting for the next security gap.
I made this mistake early on. Typed something in once, immediately knew I had done it wrong, and stopped. The AI companies have been too slow to build clean, secure ways to feed context into these systems without it going into the chat history. Grok actually has a vault feature built in — you can store credentials and share them securely. Claude does not have anything comparable yet, which is frustrating. I ended up building my own credential management layer so my bots can access what they need without any of it touching the chat log.
If your team is using any AI tools — Notion, ChatGPT, Claude, anything — this is worth a five-minute conversation. Remind them: the chat is not a password manager. If you are a Notion AI user specifically, it would be worth having the AI itself scan back through your chat history and flag anything that looks like it should be rotated.
Anyway. Good show this week. A lot to chew on.
Hope everyone has a great, profitable rest of the week. Talk soon.
— Albert
Questions about anything we covered today? Schedule a free IT consultation and we’ll walk through it with your business in mind.
Wherever your team is based, ACS provides hands-on managed IT support — see how we help businesses in Traverse City & Northern Michigan.