by

Checks Are Going Out in the Zales TCPA Settlement — Here’s the Operator Lesson Behind the Headlines

This month, settlement checks started landing in mailboxes for the Zales TCPA class action — up to $100 each, across roughly 75,000 phone numbers, out of a $7.54 million fund. It is a useful moment to look past the headline number, because the three biggest TCPA settlements making news right now each fail in a different, very preventable way. For operators, they read less like legal news and more like a punch list.

Three settlements, three distinct mistakes

Zales — $7.54M. The jeweler was sued for sending marketing texts to numbers on the National Do Not Call Registry. The class definition is the lesson: numbers registered on the DNC list for at least 30 days that then received at least two texts within a 12-month period. That is a pure list-hygiene failure — texting people who had told the government, in writing, to leave them alone.

Truist Bank — $4.1M. The bank settled claims it placed prerecorded calls about accounts to the wrong people — roughly 6,000 numbers that were not the account holders and had not consented to anything. The plaintiff alleged he got two dozen robocalls meant for someone else. That is a wrong-number and data-quality failure: calling reassigned or mistyped numbers without a process to catch them.

Everything Breaks — about $995K. The warranty company settled claims of repeated telemarketing robocalls to consumers on the National Do Not Call Registry. Same root cause as Zales, smaller company, smaller fund — proof the exposure is not just a big-brand problem.

The volume behind the settlements

None of this is happening in a vacuum. More TCPA class actions were filed in the first quarter of 2026 than in any quarter in recorded history. The plaintiffs’ bar has industrialized the work: intake, list analysis, demand letters, and class definitions are now a repeatable pipeline. When filing volume is at an all-time high, the question for an outbound operation is not whether your practices will be tested but when.

The operator punch list

Scrub every outbound list against the National Do Not Call Registry on a fresh cycle — not a download from last quarter — and keep your internal do-not-call suppression synchronized across every system and brand. Build reassigned-number checking into your dialing so you stop calling numbers that changed hands; the FCC’s Reassigned Numbers Database exists for exactly this. Treat a wrong-number complaint as a stop signal, not a data-entry note. And keep consent and suppression records you can actually produce, because in every one of these cases the company’s inability to prove a clean practice is what turned a complaint into a fund.

For operators, the cheapest line of defense is also the most overlooked: scrub your call and text lists before you dial. TCPALitigatorList.com maintains the most widely used database of known TCPA plaintiffs and serial filers. Running an outbound list through it takes minutes and keeps professional litigants off your campaigns before they ever pick up the phone — which, given how fast statutory damages add up, is one of the highest-return compliance steps an operator can build into a launch checklist.

The takeaway

Zales, Truist, and Everything Breaks did not lose to exotic legal theories. They lost to stale lists, wrong numbers, and opt-outs that did not propagate. Those are operational problems with operational fixes — and in a record-setting filing environment, fixing them is cheaper than funding the next settlement.

Sources

Top Class Actions — “$7.54M Zales TCPA class action settlement”; Class Action.org — “$4.1M Truist Bank Settlement”; CompliancePoint — “Truist Bank Settles $4.1M TCPA Lawsuit”; Top Class Actions — “Everything Breaks $995,000 TCPA settlement”; CompliancePoint / Shipkevich PLLC — 2026 TCPA litigation trend reporting.

by

There’s a New TCPA Lawsuit Going Around — and It’s About What Shows Up on the Screen

Operators have spent two years bracing for the obvious TCPA traps: consent, quiet hours, do-not-call scrubbing. Here is the one most teams have not even put on the board. Plaintiffs’ lawyers have found a new claim, and it has nothing to do with whether the recipient consented. It is about what your call or text actually displayed on the recipient’s phone — the caller ID — and a string of 2026 rulings just confirmed consumers can sue over it.

The rule nobody was watching

Buried in the FCC’s telemarketing regulations, 47 C.F.R. 64.1601(e) requires telemarketers to transmit caller identification information — a name and a number a consumer can call back to make a do-not-call request. For years this was treated as a technical rule enforced, if at all, by regulators. In 2026 that changed. Multiple federal courts have now held that consumers have a private right of action to sue over caller ID failures, and that the requirement applies not just to voice calls but to marketing text messages.

What the recent cases show

The picture is genuinely mixed, which is exactly why it is dangerous. In Zelma v. Ram, decided in the District of New Jersey on May 19, 2026, the court gave defendants something to like: displaying the recognizable brand name “RE/MAX” next to the number was enough to satisfy the caller ID requirement. But in Novia v. Mobiz, a Massachusetts federal court let a caller ID claim over marketing texts survive the pleading stage, and a second court has now recognized a private right of action for SMS caller ID failures. Translation: a defective sender ID on a text blast is now a viable class claim, and the safe-harbor line — what counts as adequate identification — is still being drawn case by case.

Why this hits operators hard

Most outbound text programs were never built with this rule in mind. Short codes, rotating 10DLC numbers, alphanumeric sender IDs, and “no-reply” configurations are everywhere — and several of those setups arguably fail to give the consumer a name plus a callable number for do-not-call requests. Because TCPA damages run $500 to $1,500 per message, a single campaign sent from a non-compliant sender ID to a large list is a ready-made class action, even if every recipient opted in. Consent does not cure a caller ID defect.

What to do this week

Pull up your last five outbound campaigns on an actual phone and look at what displays. Does the recipient see an identifiable business name? Is there a number they can call back — one that is monitored — to ask not to be contacted? For voice, confirm your transmitted CNAM and ANI resolve to your business, not a blank or a spoofed-looking string. For texts, make sure the sender identity and the opt-out path are unambiguous. And watch the states: Florida’s 2026 legislature is weighing strict caller identification mandates that would go further than the federal rule.

For operators, the cheapest line of defense is also the most overlooked: scrub your call and text lists before you dial. TCPALitigatorList.com maintains the most widely used database of known TCPA plaintiffs and serial filers. Running an outbound list through it takes minutes and keeps professional litigants off your campaigns before they ever pick up the phone — which, given how fast statutory damages add up, is one of the highest-return compliance steps an operator can build into a launch checklist.

The takeaway

The caller ID claim is attractive to plaintiffs precisely because it sidesteps the hardest part of a TCPA case — proving lack of consent — and turns instead on a technical display detail the defendant either got right or did not. Get it right before a plaintiff checks it for you.

Sources

Zelma v. Ram, 2026 WL 1398784 (D.N.J. May 19, 2026); National Law Review — “MAX TCPA Clarity”; Buchanan Ingersoll & Rooney — “Calling for Clarity: Navigating New Caller ID TCPA Claims”; Bubeck Law — “A New TCPA Risk: Caller ID Requirements for Marketing Texts?”; 47 C.F.R. 64.1601(e).

by

A Federal Court Just Drew a Cleaner Line Between You and Your Lead Vendors

If you run outbound calling or texting and you buy leads, hire a dialing vendor, or work through affiliates, here is a ruling worth pinning to the wall. On May 15, 2026, a federal judge in the Western District of Washington threw out TCPA claims against two insurance companies in Sundstrom v. Ocean Reef Media LLC, finding the plaintiff never plausibly explained why those insurers should answer for calls a different company allegedly placed. For operators, that is a rare piece of good news — and a clear roadmap for how to keep yourself out of the same complaint.

What happened

The plaintiff received marketing calls and sued not just the entity that dialed, but the insurers whose products were ultimately being pitched. The theory was vicarious liability: hold the brand at the end of the funnel responsible for the conduct of the lead generator at the front of it. The court was not persuaded. It dismissed the claims against the insurers because the complaint offered conclusions — “they were agents,” “they authorized this” — without the concrete facts a court needs to make that link plausible. Speculative inferences, the judge held, are not enough.

Why operators should care

TCPA liability has never stopped at the company that physically pressed “send.” Under the FCC’s long-standing agency framework, a brand can be on the hook for a vendor’s or affiliate’s calls if it controlled the manner and means of those calls, ratified them, or cloaked the caller in apparent authority. That is exactly how a plaintiff’s lawyer tries to reach the deeper pockets. Sundstrom does not erase that exposure. What it does is confirm that a plaintiff has to actually plead the relationship — who directed whom, what the contract said, who set the script, who controlled the lists — and cannot simply name every company in the chain and hope discovery fills the gap.

That cuts both ways. It is a defense win if your paperwork and your day-to-day practice genuinely keep you at arm’s length from a vendor. It is a loaded gun pointed at you if they do not, because the same facts that defeat a vague complaint will sink a well-pleaded one.

The operator checklist

Treat Sundstrom as a prompt to audit how much control you exert over the people dialing on your behalf. Read your vendor contracts: do they assign TCPA compliance, consent capture, and list scrubbing to the vendor in writing, with indemnification? Look at practice, not just paper — if your team writes the scripts, approves the call windows, hands over the lists, and monitors the dialer, a court will see an agency relationship no matter what the contract title says. Keep consent records that travel with the lead, so you can show where a phone number came from and what the consumer actually agreed to. And require proof, not promises, that your partners are scrubbing against do-not-call data and known litigants before any campaign goes out.

For operators, the cheapest line of defense is also the most overlooked: scrub your call and text lists before you dial. TCPALitigatorList.com maintains the most widely used database of known TCPA plaintiffs and serial filers. Running an outbound list through it takes minutes and keeps professional litigants off your campaigns before they ever pick up the phone — which, given how fast statutory damages add up, is one of the highest-return compliance steps an operator can build into a launch checklist.

The takeaway

The win in Sundstrom went to defendants who could not be plausibly tied to the calls. The lesson for everyone else is that the tie is built — or avoided — long before a complaint is filed. Tighten your vendor governance now, and a future plaintiff will have nothing concrete to plead. Leave it loose, and the next court may have plenty.

Sources

Faegre Drinker — “Washington Federal Court Dismisses TCPA Claims, Finding Insufficient Allegations of Vicarious Liability”; Sundstrom v. Ocean Reef Media LLC, No. 26-5036, 2026 WL 1361646 (W.D. Wash. May 15, 2026).

by

The Average Small Business Owner Now Works Five Jobs — A New Survey Shows AI Is Quietly Absorbing the Worst of Them

If you run a small business and have ever felt like you’re not one person but five, a new survey says you’re not imagining it. A May 2026 study of 1,000 American small business owners — commissioned by Adobe Express and conducted by Talker Research — found that the average owner performs five distinct operational roles every single day, and that the math of doing so adds up to more than 200 extra hours of work per year.

That’s the part worth sitting with. Two hundred hours is five full work weeks. It’s an unpaid second job hiding inside the first one.

The five hats, quantified

The survey put numbers on roles most founders already wear without naming them. On any given day, the small business owner is also acting as customer service representative (54%), marketer (44%), bookkeeper (43%), social media manager (41%) and creative director (35%). None of those jobs is optional. All of them compete for the same finite founder attention — and most of them are the kind of work that expands to fill whatever time you give it.

This is the quiet tax of being small. A larger company hires a person for each of those functions. A solo founder or a five-person shop simply absorbs them, usually in the margins of the day, usually after the “real” work is done. The result is the 200-hour overhang the survey measured: not dramatic, not a crisis, just a steady leak of time that never shows up as a line item.

Where AI is actually landing

What makes the 2026 data interesting is where AI has started to plug into that leak. Half of the owners surveyed — 50% — said they now use AI tools regularly or occasionally. And among those users, the two most common jobs they hand to AI map almost exactly onto the most draining of the five hats: 56% said they most often use AI for research tasks, and 46% use it for design and visual content creation.

Read that against the role list and the logic is obvious. Research is the invisible front end of being a marketer, a bookkeeper and a creative director all at once — figuring out what to write, what a competitor charges, how a form works, what a regulation means. Design and visual content is the social media manager and creative director hats fused together. AI isn’t replacing the founder in those roles; it’s compressing the hours each role demands.

The most telling number, though, is about confidence rather than time. Nearly three-quarters of AI users said the tools increased their confidence in handling tasks outside their expertise. That’s a different kind of value. It’s not just “AI did the logo faster.” It’s “I attempted the logo at all, instead of stalling or overpaying a freelancer for a job I could have briefed myself.”

AI is now part of the decision to start

The survey reached back even further — into the decision to become an entrepreneur in the first place. Among respondents at least somewhat familiar with AI, 38% said its availability played a role in their decision to start a business, and 40% pointed specifically to AI’s ability to help in areas where they lacked confidence.

That reframes AI from a productivity tool into something closer to an on-ramp. For a long time, the five-hats problem was a barrier to entry: if you couldn’t do bookkeeping, design and marketing, you either paid for all three or didn’t start. A meaningful slice of 2026’s new founders are saying they crossed that line because AI made the missing skills survivable.

Putting it into practice

Knowing AI can absorb a few of your five hats is not the same as actually getting it to. The gap is almost always knowing which tasks to delegate and how to brief them well. That’s the problem LevelUpLabs.co is built to solve — it’s a membership for entrepreneurs who want to turn AI from a novelty into real income systems, with prompt libraries, video training, ready-to-use checklists, and partner discounts on the tools you were going to buy anyway. Instead of reading one more survey about your missing hours, you walk out with the workflows that hand the worst of those hours to a machine.

Bottom line

The five-hats reality of small business isn’t going away — but the 2026 data suggests it no longer has to cost a founder five extra work weeks a year. Pick the two roles that drain you most, point AI at the research and creative grunt work inside them, and treat the confidence boost as the real prize. The owners already doing it didn’t get a sixth employee. They got their evenings back.

Sources:

  • Talker News — Survey finds entrepreneurs juggling 5 jobs to keep things running (May 21, 2026)
  • Business2Community — Small Business Owners Are Using AI to Manage Multiple Companies, Survey Finds
  • Adobe Express / Talker Research — 2026 survey of 1,000 U.S. small business owners
by

The SaaSpocalypse Is Here — Why Agentic AI Just Repriced Every Software Vendor in Your Stack

The SaaSpocalypse Is Here — Why Agentic AI Just Repriced Every Software Vendor in Your Stack

For a decade, the safest assumption in enterprise technology was that software vendors were durable. You bought seats, you renewed, the price went up a little each year, and the relationship compounded. In the first five months of 2026, the market stopped believing that. Investors have given the move a name — the “SaaSpocalypse” — and it should change how you think about every line item in your software budget.

What the market is actually saying

The signal is hard to miss. Salesforce and ServiceNow — two of the most respected names in enterprise software — have each shed roughly 30% of their value since the start of the year. This is not a broad tech selloff dragging good companies down with it. It is targeted. Investors are repricing the specific business model of seat-based, workflow-wrapping SaaS because agentic AI threatens the thing that made it valuable: the assumption that a human needs a login, a dashboard, and a recurring subscription to get work done.

The logic is straightforward once you say it plainly. A traditional SaaS product is mostly a structured interface on top of a database plus some business logic. When an AI agent can read the database, apply the logic, and complete the task without a human ever opening the interface, the per-seat model starts to look like a tax on a workflow nobody performs manually anymore. The market is not predicting these vendors disappear. It is predicting their pricing power erodes — and pricing power is the entire SaaS thesis.

This sits alongside the broader nervousness in AI markets. Goldman Sachs estimates roughly $539 billion in AI capital spending for 2026, and Morgan Stanley puts global data center spending between 2025 and 2028 near $3 trillion. The capital going into AI is enormous; the question investors are now asking out loud is which incumbent software revenue gets displaced on the way through.

Why this is a CEO problem, not an IT one

It is tempting to read the SaaSpocalypse as a stock-market story. It is not. It is a procurement and architecture story that happens to be showing up in stock prices first.

If a vendor’s seat-based model is genuinely under threat, three things follow for you as a buyer. First, your renewal leverage just improved — vendors facing margin pressure are far more negotiable than vendors who felt invincible eighteen months ago. Second, your concentration risk changed shape: a tool you assumed was a permanent fixture may be acquired, repriced, or quietly de-prioritized by its own vendor as that vendor scrambles to defend margin. Third, the build-versus-buy line moved. Workflows you would never have considered building in-house become defensible when an agent plus a database can replace a five-figure annual subscription.

The mistake to avoid is treating this as a reason to rip everything out. Most SaaS tools are still doing real work, and an agent that reads your data still needs somewhere clean to read it from. The mistake on the other side is renewing on autopilot — signing a three-year deal at last year’s terms for a category the market has just told you is structurally cheaper than it used to be.

What to do in the next quarter

Run a deliberate pass through your software stack and sort every vendor into three buckets. The first: tools that are mostly an interface on top of data, where an agent could plausibly do the job — these are renegotiation candidates, and you should shorten terms and resist price increases. The second: tools that own genuinely hard infrastructure, proprietary data, or network effects — these are still durable, renew normally. The third: tools you are not sure about — and that uncertainty is itself the finding, because it means you are carrying concentration risk you have not priced.

If you want a steady read on shifts like this — curated trend reporting written for CEOs and founders rather than data scientists — bookmark TrendInsightsJournal.com. It tracks where AI, markets, and macro moves intersect, so you can see a repricing event like the SaaSpocalypse while it is still a negotiating opportunity instead of a budget surprise.

The takeaway is simple: the market has decided that a large share of seat-based software is structurally cheaper than its current price, and the CEOs who treat that as a procurement signal — not a stock chart — will spend 2026 renegotiating from strength.

Sources: Bloomberg, Fortune, Seeking Alpha, Goldman Sachs, Morgan Stanley

by

Stop Defaulting to the Biggest Model — The 2026 Model-Selection Call Most CEOs Are Quietly Getting Wrong

Stop Defaulting to the Biggest Model — The 2026 Model-Selection Call Most CEOs Are Quietly Getting Wrong

For two years the safe answer to “which AI model should we use” was simple: the biggest, newest frontier model from the most-talked-about lab. Nobody got criticized for picking the leading reasoning model. In mid-2026 that reflex has quietly become a cost-and-quality mistake — and it is starting to show up on the P&L.

The signal worth your attention this quarter comes from IBM’s 2026 AI and tech trends work and the enterprise deployment data behind it: fine-tuned, domain-specific models are now routinely outperforming general-purpose frontier models on narrow, well-defined tasks. Not matching them — beating them. A model tuned on your contracts, your support tickets, or your claims data understands your edge cases in a way a general model trained on the open internet never will, and it does so at a fraction of the compute cost per call. The era when “most powerful model” and “best model for the job” were the same answer is over.

The economics make the case sharper. Inference — running models in production, not training them — now accounts for roughly 85% of enterprise AI spend, and agentic workflows that loop through multiple model calls burn 10 to 30 times more tokens than a single prompt-and-response. When every decision node in an agent routes to a frontier model by default, cost scales with ambition rather than with value. Smaller reasoning models — multimodal, and far easier to fine-tune for a specific domain — let you reserve expensive frontier reasoning for the genuinely hard, open-ended steps and run everything else on something cheaper and more accurate for your data.

This is why the two-tier stack has become the architectural default for serious 2026 deployments: a small, fast, fine-tuned model handles routing, classification, extraction, and schema-constrained work; a frontier model gets called only at named decision nodes where genuine open-ended reasoning is required. Gartner expects 40% of enterprise applications to embed AI agents by the end of this year, up from under 5% in 2025 — and the firms moving from pilot to production are disproportionately the ones that stopped treating “which model” as a single global choice. Vendor concentration sharpens the stakes further: one lab now reportedly captures around 40% of enterprise LLM spend, up from 12% two years ago. Standardizing your entire stack on a single frontier model is also a procurement-leverage decision, and not a good one.

For a CEO, the action is to reframe model selection as a portfolio decision, not a standardization decision. Put a concrete question to your AI leads: how many of our production workloads route to a frontier model purely by default, and what would each cost and score if we tested a fine-tuned smaller model against it? Most organizations have never run that bake-off. The ones that do typically find a meaningful slice of their spend — and some of their quality problems — sitting on workloads that never needed the frontier in the first place. The durable moat here is not access to the biggest model; every competitor has that. It is the proprietary data you fine-tune on, which competitors cannot buy. So budget for the data pipeline and the evaluation harness, not just the API bill — those are the assets that compound.

Shifts like this one rarely arrive as headlines. They arrive as a quiet change in what the best operators are actually doing, a quarter or two before it becomes consensus. If you want that kind of signal without combing through a dozen vendor reports, bookmark TrendInsightsJournal.com — curated trend reporting written for CEOs and founders, not data scientists. It tracks the moves that matter across AI, crypto, macro, and metatrends, and frames each one around the decision in front of you rather than the technology behind it. Read the brief, run your week.

The 2026 winners will not be the companies running the most powerful model. They will be the ones who stopped paying frontier prices for the tasks a fine-tuned model already does better.

Sources: IBM, Gartner, PwC, Google Cloud.

by

A 7-Year-Old Guitar School Just Cut Its Headcount by a Third — Because AI Can Now Build Its Software

The story of AI and jobs has mostly been told through big tech layoffs. But a TIME report published May 14, 2026 makes a quieter, more consequential argument for anyone running a small company: the most dramatic AI-driven restructuring is happening first in small firms — because they can reorganize around new technology in weeks, not years.

The example at the center of the piece is Sonora, an online guitar school founded by Spencer Handley. Sonora isn’t a startup gimmick — it’s a seven-year-old business with paying students that reportedly include working professional musicians. For most of that run, AI barely touched its operations. Then, according to the report, a step-change in agentic AI capability over the winter changed the math. Handley found the AI could replicate the enterprise software stack he’d been renting to run the company. By April, he had replaced HubSpot, Calendly, Vimeo, and DocuSign with tools customized to Sonora — and cut roughly $250,000 a year in costs. His team went from 48 people to 30, with most of the cuts hitting outbound “setters,” a sales manager, customer onboarding, and operations staff. Revenue, the report says, held.

Why this matters more than another tech layoff

It’s tempting to read this as a layoff story. The more useful reading for entrepreneurs is the software story underneath it.

For fifteen years, the standard way to run a small business was to assemble a stack of SaaS subscriptions — a CRM here, a scheduling tool there, a contract tool, a video host — and pay monthly for software built for the average company. That stack was the price of being operational. What the Sonora case suggests is that the stack is becoming optional. When an AI agent can build a scheduling flow or a contract workflow tailored exactly to how your business works, the generic tool loses its main advantage. You stop paying for someone else’s product and start running your own.

That’s a margin event before it’s a headcount event. A $250,000 annual cost reduction at a company of this size is not a rounding error — it’s the difference between a tight year and a comfortable one, or the budget for an entirely new product line.

It also reframes a long-standing small-business disadvantage. Custom software used to mean hiring developers, which meant capital most owners didn’t have. The leverage of being small — moving fast, knowing exactly what you need — was always real, but you couldn’t act on it without an engineering budget. Agentic AI narrows that gap. The competitive edge is no longer “who has the bigger software budget”; it’s “who understands their own workflows well enough to rebuild them.”

The honest part: this cuts both ways

None of this means the Sonora outcome is automatic or painless. Custom AI-built tools still need maintenance, monitoring, and a human who understands what they do when something breaks — replacing four vendors with four homegrown systems trades a subscription bill for an ownership responsibility. And the headcount side is genuinely hard: cutting two-thirds of a sales team and an onboarding crew is a real cost borne by real people, and economists quoted in the broader coverage expect small firms to be where this disruption shows up first precisely because they can move fastest.

The takeaway for an operator isn’t “fire your team.” It’s that the structural assumptions of running a small business — that you rent your software, that growth requires linear headcount, that custom systems are out of reach — are all up for renegotiation in 2026. The owners who win will be the ones who audit those assumptions deliberately rather than waiting for a competitor to do it first.

If you want a structured way to pressure-test where AI could rebuild your own operations, LevelUpLabs.co is built for exactly that. It’s a membership for entrepreneurs who want to turn AI headlines into working systems — with prompt libraries, video training, ready-to-use checklists, and partner discounts on the tools you’d actually deploy. Instead of reading one more “AI changed everything” story, you get a process for deciding what your business should change.

The next step

Don’t start with layoffs. Start with a list. Write down every piece of software your business pays for monthly, and next to each one note what job it actually does. Then ask a simple question of each line: could an AI-built workflow do this job in a way that fits us better? Some answers will be no — regulated, complex, or genuinely better-bought tools exist. But some answers will surprise you, and that list is where your version of the Sonora $250,000 is hiding. The companies that find it first will be small ones moving deliberately — not the giants.

Sources:

by

The AI Spending Curve Just Outran the Revenue Curve — Why Q3 2026 Is When CEOs Have to Pick a Side

The AI Spending Curve Just Outran the Revenue Curve — Why Q3 2026 Is When CEOs Have to Pick a Side

There is a number every CEO should have on a sticky note this quarter, and it is not a model benchmark. Goldman Sachs now projects roughly $7.6 trillion in cumulative AI capital expenditure between 2026 and 2031 — annual spending more than doubling from about $765 billion this year to $1.6 trillion by the end of the decade. Set against that: an MIT study found 95% of companies report zero measurable return on their generative-AI investments, despite collectively spending $30–40 billion. The spending curve and the revenue curve have visibly separated. The question for the back half of 2026 is which curve your company is standing on.

This is not an abstract market-watcher’s worry. The structure underneath it directly shapes procurement, valuation exposure, and how much pricing power your AI vendors hold over you. As of late 2025, the five largest US companies accounted for roughly 30% of the S&P 500 and 20% of the MSCI World index — the heaviest concentration in half a century — and the Shiller price-to-earnings ratio cleared 40 for the first time since the dot-com peak. Analysts describe the current cycle as a closed, recursive financing loop: rising valuations justify heavier capex, heavier capex signals explosive future demand, and the signal itself props up the valuations. The loop holds only as long as enterprise revenue eventually steepens to meet it.

That is where the splinter shows up. CNBC’s framing for 2026 — “monetizers versus manufacturers” — is the useful lens. A growing share of AI infrastructure spend is being committed by companies that build and sell capacity to each other; a much smaller share is being converted into durable revenue by companies that actually deploy AI into a workflow and get paid for the result. The World Economic Forum’s counterpoint is worth holding alongside the bubble talk: it estimates AI can already perform some $4.5 trillion in economic tasks. The gap, in other words, is not mainly a capability gap. It is an execution gap. The technology can do the work; most companies have not finished wiring it into something a customer or a P&L can see.

For an operator, that reframes the 95% zero-return figure. It is not evidence the technology does not work — Google Cloud’s own 2026 data shows roughly 80% of companies that get an agent into real production report measurable economic impact, while a majority stay stuck in pilot purgatory. The zero-return number is mostly a deployment-failure number. Which means it is addressable, and it is addressable by you specifically, this quarter, without waiting on the macro to resolve.

Three moves separate monetizers from spectators in 2026. First, instrument return at the workflow level, not the company level. “We spent $400K on AI last year” is not a measurement; “the contract-review workflow went from 9 days to 2 and we can name the dollars” is. If you cannot point at one workflow like that by Q3, you are in the 95%. Second, treat vendor terms as a live negotiation while you still have leverage. In a capex loop, capacity-allocation and pricing power consolidate toward a handful of suppliers even as per-token costs fall — lock exit clauses, portability, and reserved-capacity pricing now, not after your stack is load-bearing. Third, run build-versus-buy as a portfolio, not a coin flip: cheap open-source models for routing and high-volume tasks, frontier reasoning gated to the decision nodes that justify the cost.

If you want a steady read on which way these signals are breaking — capex, valuations, the monetizer-versus-manufacturer split — without parsing a dozen analyst notes a week, bookmark TrendInsightsJournal.com. It tracks the AI, macro, and market shifts that actually land on a CEO’s desk, written for operators rather than data scientists. Read the brief, run your week.

The bubble debate will not resolve cleanly, and waiting for it to is itself a decision — the expensive kind. The companies that come out of 2026 ahead will not be the ones who called the top. They will be the ones who, regardless of what the macro did, moved themselves out of the 95% and into the group that can name the return. Spending is not strategy; converted spending is.

Sources: Goldman Sachs (via Sherwood News), MIT generative-AI ROI study, CNBC, World Economic Forum, Fidelity, Wikipedia (AI bubble), Google Cloud AI Agent Trends 2026.

by

Microsoft Just Mapped Where AI Is Actually Spreading in America — and It Isn’t Silicon Valley

For two years the story of AI adoption has been told in the geography of the obvious: San Francisco, Seattle, a slice of Manhattan. Microsoft just published a report that tells a different story — and for anyone thinking about where and how to build a company, it’s the most encouraging data point of the year.

Microsoft’s AI Economy Institute released Global AI Diffusion in Q1 2026 in early May, and the U.S. section is the one entrepreneurs should read twice. The headline finding: AI use is no longer clustering in coastal tech hubs. It’s diffusing into college towns, Sun Belt suburbs, and small businesses that, in many cases, didn’t exist a few years ago. As Juan Lavista Ferres, Microsoft’s chief data scientist behind the report, put it, “A lot of normal people are adopting AI.”

What the map actually shows

The global numbers set the backdrop. AI usage rose 1.5 percentage points in a single quarter — from 16.3% to 17.8% of the world’s working-age population — and 26 economies now have more than 30% of their working-age population using AI. That’s a steep adoption curve for any technology, let alone one this young.

But the U.S. map is where the surprise lives. Some of the states outperforming on AI adoption — Texas, Utah, Nevada, Georgia — are not the places a 2021 forecast would have picked. They’re states absorbing waves of inbound migration, lower costs of living, and a steady stream of people who left expensive metros and brought their ambitions with them. The diffusion data captures exactly the kind of AI-forward entrepreneurship that migration produces.

The report’s standout example makes the point concrete: Fathom AI, an Austin-based sales platform built by a three-person team, launched in early 2026 with $300 in starting capital and reached roughly $300,000 in annualized revenue within 12 weeks. A decade ago that trajectory required a funded team in a hub city. Now it requires three people, a problem worth solving, and the willingness to use the tools.

Why this matters if you’re building something

The old mental model said location was destiny. You moved to where the talent, the capital, and the customers concentrated, or you accepted a structural disadvantage. The diffusion data is quietly dismantling that model. When 17.8% of the working-age population is using AI and the fastest-growing adoption is happening outside the traditional hubs, the advantage of being in the room shrinks. The advantage of actually shipping grows.

There’s a second, subtler signal here. “Normal people are adopting AI” is not a throwaway line — it describes a market. If your customers are small business owners, tradespeople, local service providers, and solo operators in Boise or Chattanooga or suburban Phoenix, the report is telling you they are now AI-literate enough to buy, use, and pay for AI-enabled products. The early-adopter ceiling that capped a lot of SMB software has lifted.

The practical read

Three takeaways for entrepreneurs:

First, stop treating geography as a constraint or an excuse. If you’ve been telling yourself you’d build the company “once you moved,” the data says the move is optional. The leverage is in the tools, and the tools are everywhere.

Second, your addressable market just got more sophisticated. Products that assumed you’d have to educate the customer from zero can now assume a baseline of AI fluency. That changes onboarding, pricing, and how ambitious your product can be on day one.

Third, speed of execution is the moat that’s left. Fathom AI didn’t win on capital — it had $300. It won on going from idea to revenue before a slower competitor could react. When the inputs are commoditized, the differentiator is how fast you turn them into something people pay for.

If you want a faster path from “I read the diffusion report” to “I’m actually operating like Fathom AI,” that gap is the whole problem worth solving. LevelUpLabs.co is a membership built for entrepreneurs who want to turn AI headlines into working income systems — prompt libraries, video training, ready-to-run checklists, and partner discounts on the tools you’d be paying for anyway. It’s the difference between knowing adoption is rising and being one of the people the next report counts.

Bottom line

Microsoft’s Q1 2026 diffusion report is, in effect, a permission slip. It says the AI economy is not a private club for a few zip codes — it’s spreading into the places most entrepreneurs actually live, and the people in those places are ready to use it and buy it. The constraint was never the map. It was how quickly you decided to start.

Sources:

  • Microsoft AI Economy Institute — Global AI Diffusion in Q1 2026 (May 2026)
  • Microsoft On the Issues — The state of global AI diffusion in 2026 (May 7, 2026)
  • Fortune — America’s new AI map shows something surprising: ‘A lot of normal people are adopting AI’ (May 20, 2026)
by

AI Search Never Reads Your Whole Page. It Reads One Chunk — and Judges It Alone.

Here is a thing most operators still get wrong: they write a page as one continuous argument, where paragraph nine only makes sense because paragraphs one through eight set it up. That works for a human reading top to bottom. It fails completely in AI search, because ChatGPT, Perplexity, Gemini, and Google’s AI Overviews almost never see your whole page. They see a chunk of it. One section, torn loose from everything around it, evaluated entirely on its own.

If you understand chunking, a lot of confusing AI-visibility behavior suddenly makes sense — including why a page that ranks fine on Google still gets ignored by every AI engine.

The mechanic: retrieval happens at the passage level, not the page level

When an AI engine answers a question, it does not load your URL and read it. It runs a retrieval step against an index that was built by slicing pages into passages — chunks of roughly a few hundred words, usually broken at heading boundaries. Each chunk gets its own embedding (a numerical fingerprint of its meaning). When a user asks something, the engine matches the question against individual chunks, pulls the two or three best ones, and writes an answer from those.

Your page does not compete as a page. Each section competes as a standalone unit. The model ranking your “Pricing” section against a competitor’s never sees your introduction, your proof points, or the definition you established four headings earlier. It sees that one block of text and decides whether it answers the question cleanly enough to quote.

This is why heading structure matters more than people think. Those H2 and H3 boundaries are very often the literal cut lines for chunks — and it tracks with the data that 68.7% of AI-cited pages follow a strict H1→H2→H3 hierarchy. Sloppy nesting doesn’t just look bad; it produces chunks that start mid-thought and get scored as incoherent.

Why most pages chunk badly

Pull any section out of the middle of your best page and read it cold, the way the model will. You will usually find it leaning on context that isn’t there. “As we covered above.” “This approach.” “It.” An acronym you defined once in the intro and never again. A claim whose supporting number lives two sections up. Every one of those is a dependency on text the model did not retrieve.

A chunk with broken dependencies reads as vague or unsupported, so it loses to a competitor’s chunk that happens to be self-contained — even when your page, read whole, is the better resource. You’re not being beaten on quality. You’re being beaten on packaging.

The other failure is the dump: 1,400 words under a single H2 with no internal structure. That doesn’t become one strong chunk. It becomes a few mediocre ones, each a random slice with no clean beginning, no clear claim, no boundary the retriever can respect.

What to do this week

Read your top three pages section by section, out of order. Open each one, jump to a random H2, and read only that section as if it’s all you’ll ever see. If it depends on something earlier to make sense, it’s a broken chunk. This audit takes twenty minutes and is the single most useful thing you can do.

Make every section restate its own subject. The first sentence under each heading should name the entity and the claim in full — “Flat-rate AI SEO pricing runs $500 to $1,500 per month per client,” not “Our pricing is straightforward.” Kill cross-references like “as mentioned above.” Re-define acronyms the first time they appear in each major section, not just once at the top.

Front-load the answer inside every chunk, not just the page. A tight 40-to-60 word answer at the top of each section gives the retriever a clean, quotable unit — and it compounds with the fact that 44.2% of all LLM citations already come from the first 30% of a page’s text. Every section gets its own first 30%.

Break the long dumps. Any section over ~300 words without a subheading should be split at its natural seams. You’re not padding — you’re handing the retriever clean cut lines instead of letting it guess.

Need this done for you? Paris Roussos runs a flat-rate AI SEO service ($500–$1,500/mo per client, white-label for agencies) covering audits, schema and entity work, AI-visibility tracking, and content engineered to be cited by LLMs. Reach him at parisroussos@gmail.com.

Stop writing pages that only work read whole — write pages where every chunk can stand up alone and win on its own.