The Punch List: April 2026
By Tim Dini | April 2026 | Issue #2
Month two. Still here.
Last month I launched this thing and half-expected nobody to notice. Turns out a few of you did, which means I now have to do it again. Good. Accountability is underrated.
This month’s list is shorter than the first one. Six items instead of eight. That’s not because March was slow. It’s because I got pickier. I spent the month watching people forward me AI tool recommendations with subject lines like “This changes EVERYTHING” and I kept asking the same question: would a plumber in Phoenix care about this? Would a dentist in Charlotte? If the answer was “only if they also happen to be a software engineer,” I cut it.
What survived: Google is quietly building the plumbing (yes, that kind) for AI to buy things on your behalf. The FTC just banned an AI company for lying to small businesses. A new category of tools is emerging to track whether AI even knows your business exists. The company that makes the AI I use to research this newsletter accidentally leaked its most powerful model through a security mistake that would be funny if it weren’t terrifying. A vendor’s sales pitch accidentally uncovered something your team is probably doing behind your back. And Burger King put AI in their employees’ headsets, which is the smartest thing they’ve done in a while, because it sure wasn’t the Whopper “makeover.”
Let’s get to it.
The Punch List
Google is building a checkout inside AI search
Universal Commerce Protocol (UCP): An open standard that lets AI agents handle shopping from discovery to purchase.
https://ucp.dev/
Google announced UCP at the National Retail Federation conference back in January, but the March 19 update is where it starts getting real. They added multi-item cart support (so an AI agent can build a full shopping cart, not just buy one thing), real-time catalog access for live pricing and inventory, and identity linking so shoppers keep their loyalty perks when buying through AI. Salesforce, Stripe, and Shopify are all building it into their platforms. The list of companies that endorsed this thing reads like a retail hall of fame: Walmart, Target, Wayfair, Best Buy, Home Depot, Visa, Mastercard, American Express.
Here’s what this actually means. Right now, when someone searches in Google’s AI Mode or asks Gemini about a product, they get a recommendation and then click through to buy it on the retailer’s site. UCP is the infrastructure that makes the click unnecessary. The AI agent finds the product, checks real-time inventory, applies your loyalty discount, and completes the purchase. You never leave the conversation.
Google is also simplifying the onboarding process through Merchant Center, which means this isn’t just for enterprise retailers with engineering teams. If you’re on Shopify, your platform is going to support this whether you’re paying attention or not.
My Thoughts: If you sell products online, this is not optional reading.
If you’re a service business (plumber, lawyer, dentist), you’re not directly affected today. But pay attention to the trajectory. Google is training consumers to let AI handle buying decisions. Right now that’s “buy me running shoes.” At some point it’s “find me a plumber who can fix a leaking water heater tomorrow morning and book the appointment.” The behavioral shift is the story here, not the protocol itself. When your customers get comfortable letting AI buy things for them, they’re going to expect AI to book services for them too. That’s not a five-year prediction. That’s a two-year one, maybe less.
Source: Google blog announcement, March 19, 2026
2. The FTC just banned an AI company for lying to small businesses
Air AI / FTC Settlement: $18 million judgment for misleading entrepreneurs about AI earnings potential
Here’s what Air AI was selling: a conversational AI product that it claimed could replace human customer service reps and, combined with their coaching services, make business owners large sums of money. They told customers they’d earn back tens of thousands of dollars within 30 days. The FTC’s complaint, filed in August 2025, alleged the company had been running this playbook since at least February 2023, racking up roughly $19 million in consumer losses.
The settlement, announced March 24, 2026, bans Air AI’s operators from selling or marketing any business opportunity going forward. They also can’t make unsubstantiated claims while telemarketing or selling any goods and services. The $18 million judgment is largely symbolic since the operators apparently can’t pay it. They’ll pay $50,000 for consumer relief instead.
My Thoughts: I’m putting this in the Punch List because I know what your inbox looks like. If you own a business that generates leads from the internet, you are getting pitched AI tools constantly. “This AI will replace your receptionist.” “This AI will book appointments while you sleep.” “This AI will 10x your revenue in 90 days.”
Some of those tools are legitimate. Some of them are Air AI.
Here’s a simple filter: if an AI tool company leads with earnings claims instead of product demonstrations, that’s a red flag. If they promise specific dollar returns within a specific timeframe, that’s a red flag. If the refund guarantee gets conveniently hard to actually use, that’s a red flag. None of this is new advice. The FTC has been saying this about business opportunities for decades. It’s just that slapping “AI” on the pitch makes it sound more believable than it should.
The broader pattern matters too. The FTC is calling this “AI washing,” which is basically companies exaggerating what their AI can do. Expect more enforcement actions like this. The hype-to-reality gap in AI marketing is enormous right now, and regulators are starting to close it.
Source: FTC press release, March 24, 2026
3. There are now tools built specifically to track whether AI knows you exist
AI visibility monitoring: a new product category for tracking brand citations in AI-generated answers
This isn’t about one specific tool. It’s about a category that didn’t exist six months ago and is now crowded enough to write about.
The problem these tools are trying to solve is real: traditional SEO rank trackers tell you where you show up on Google’s blue links, but they tell you nothing about whether ChatGPT, Gemini, Perplexity, or Claude mention your business when someone asks a relevant question. You could be ranking #1 on Google and completely invisible in AI-generated answers.
Several platforms have launched in the last few months specifically to track “AI visibility.” They scan major AI systems with real prompts, measure how often your brand gets cited, compare your share of voice against competitors, and some of them auto-generate content designed to improve your AI citation rate. Pricing ranges from around $89/month to $500+/month depending on how many brands and AI systems you want to monitor.
My Thoughts: The problem is legitimate. AI search is eating traditional search traffic, and most businesses have no idea whether they’re being cited, ignored, or misrepresented. That part is real and it matters.
The tools themselves? I’m cautious. Most of them launched in the last few weeks. Independent reviews are basically nonexistent. The ones I’ve looked at combine AI visibility tracking with auto-generated content publishing, which feels like two very different value propositions jammed together. And some of the claims about what drives AI citations are still more theory than proven methodology.
Here’s what I’d actually recommend right now: before you spend $89-$500/month on a new tool, do the free version yourself. Open ChatGPT, Claude, Perplexity, and Gemini. Ask them the questions your customers would ask. “Who’s the best plumber in [your city]?” “What should I look for in a personal injury lawyer?” “How much does a dental implant cost?” See if your business comes up. See what your competitors are doing that you’re not. That ten-minute exercise will tell you more than any dashboard.
The category is worth watching. Semrush and Ahrefs are both adding AI visibility features to their existing platforms, which might be the smarter play if you’re already paying for one of those. But for now, I’d hold off on the new standalone tools until they have enough track record to evaluate honestly. This site doesn’t recommend things I can’t verify.
If AI visibility is new to you, the AEO Guide covers the fundamentals of how AI search finds and cites businesses.
4. The AI I use to write this newsletter just accidentally leaked its most powerful model
Claude Mythos / Capybara: Anthropic’s next-generation model, confirmed after an accidental data leak
Full disclosure: I use Anthropic’s Claude to help research and draft this newsletter. So when Anthropic accidentally left nearly 3,000 unpublished files sitting in a publicly accessible data store, including a draft blog post announcing their most powerful model ever, I paid attention.
Here’s what happened. Someone at Anthropic forgot to flip a privacy toggle on their content management system. Assets uploaded to the CMS default to public unless you explicitly change the setting. A cybersecurity researcher at Cambridge and a senior AI security researcher at LayerX found the exposed files. Among them: a draft blog post describing a new model called Claude Mythos, with a new tier called “Capybara” that sits above their current top-of-the-line Opus models.
Anthropic confirmed it’s real. A spokesperson called it “a step change” in AI performance and “the most capable we’ve built to date.” The leaked draft says Capybara scores dramatically higher than the current best model on tests of software coding, academic reasoning, and cybersecurity.
That last one is the reason this matters beyond the tech industry. The draft warned that the model poses “unprecedented cybersecurity risks” and can exploit software vulnerabilities in ways that “far outpace the efforts of defenders.” Anthropic plans to release it first to cybersecurity defense teams, which tells you how seriously they’re taking the dual-use problem. Cybersecurity stocks dropped on the news.
My Thoughts: Two things here.
First, the irony. Anthropic is the company that positions itself as the safety-first AI lab. They refused to let the Pentagon use Claude for autonomous weapons. They publish research on AI risks. And they just leaked their most sensitive product information because someone didn’t change a default setting on a content management system. If there’s a better argument for why operational security matters at every level of a business, not just the technical stuff, I haven’t seen one. Your business probably has default settings that are quietly exposing more than you think, too.
Second, the substance. The AI tools you’re using right now are about to get significantly more capable. Not in a “new button on the dashboard” way. In a “step change” way. If you’re still treating AI like a novelty or a search engine replacement, this is another signal that you’re running out of time to take it seriously. The model that helped me research three of the items in this newsletter is apparently the weaker sibling of what’s coming next.
Source: Fortune exclusive, March 27, 2026
5. Your Employees Are Already Using AI. Do You Know How?
Wrike (a project management platform) published a piece on SmartBrief last week with a headline that sounds urgent: “2026 will expose broken marketing workflows being masked by AI.” Let me save you the click. It’s mostly a sales pitch for Wrike’s own platform, built around their own survey data. Shocking, I know.
But here’s the thing. Buried inside the sales pitch are two stats from their research that local business owners need to see.
First: 42% of employees have used AI tools that their employer didn’t approve. Not “considered using.” Used. Second: only 23% of employees feel aligned with their leadership on AI strategy.
Read those together and the picture gets uncomfortable fast.
My Thoughts: Think about your front desk. Your office manager. Your associate attorneys. Your hygienists. Right now, almost half of them may be using ChatGPT, Gemini, or some other AI tool to handle tasks, and you have no idea what they’re telling it, what it’s telling them, or whether any of it matches how you actually want your business to communicate.
That’s not an AI problem. That’s a management problem. And the article’s core point (once you strip away the product pitch) is actually right: AI doesn’t fix broken systems. It makes the broken parts move faster.
For a solo plumber or a three-person insurance office, this isn’t about “workflow orchestration” or “connected intelligence” (Wrike’s words, not mine). It’s simpler than that. It’s about having a 15-minute conversation with your team: Are you using AI? For what? Show me. And then deciding together what’s helpful, what’s risky, and what needs guardrails.
You don’t need a platform for that. You need a Tuesday morning meeting.
6. Burger King put AI in their employees’ headsets (and it’s smarter than their Whopper “makeover”)
BK Assistant: AI-powered operational coaching in 500 U.S. restaurants
Burger King is piloting an AI-powered system called BK Assistant in roughly 500 U.S. restaurants. The system integrates a voice-enabled AI assistant named “Patty” directly into employee headsets. Powered by OpenAI’s technology, Patty provides real-time operational support: answering menu preparation questions, alerting managers when inventory runs low, updating digital menus when items sell out, and tracking service phrases like “welcome,” “please,” and “thank you.”
The company has been clear that the goal is coaching and operational support, not employee scoring or surveillance.
My Thoughts: I almost didn’t include this because Burger King isn’t in the audience profile for this newsletter. But the pattern is what matters, not the company.
Think about what’s actually happening here. AI isn’t replacing the employees. It’s making them better at their jobs in real time. It’s answering questions they’d normally have to ask a manager. It’s catching inventory problems before they become customer complaints. It’s doing the operational monitoring that a really good shift supervisor does, except it never gets tired and it’s working across 500 locations simultaneously.
Now think about your business. If you run a service company with field technicians, imagine every new hire having an AI assistant in their ear that knows your pricing, your procedures, and your inventory. If you run a law firm, imagine every paralegal having instant access to an AI that can answer procedural questions without pulling a senior attorney away from billable work. If you run a dental practice, imagine your front desk staff having real-time support for insurance questions and scheduling optimization.
This is where AI in operations is headed. Not replacing people. Making the people you have more capable, faster. The businesses that figure this out first are going to have a serious hiring and retention advantage.
Side Note: BK also rolled out a Whopper “makeover” recently. I tried one. It tastes like the same sandwich. They changed the bun so it doesn’t arrive looking like a pot holder, and they wrapped it in a folded paper sleeve like something Starbucks dreamed up. I think the idea is to make it a one-hand-hold-while-driving situation. But this is a Whopper we’re talking about. Long onion ‘strings’ you can’t pull far enough without sauce slapping you on the chin. Lettuce that slides out the sides like it’s making a prison break. No amount of origami wrapper engineering is going to fix that.
That’s the List
Six items, zero filler. If March was the “here’s why AI matters” issue, this one is “here’s what’s actually happening right now.” Google is building transaction infrastructure inside AI search. The FTC is punishing companies that lie about what AI can do. New tools are trying to make AI visibility measurable. The AI company I rely on every day just accidentally proved that even the experts get basic security wrong. A vendor’s survey accidentally revealed that your team is probably already using AI whether you know it or not. And a fast food chain is showing everyone what operational AI looks like when it’s done right, even if they can’t figure out a sandwich.
If something on this list saved you from wasting money on the wrong tool or made you think differently about where your business fits in all of this, that’s why I write it. If not, reply and tell me what you actually need to know. I read every one.
