The Majority Already Uses AI — and the Old Playbook Is a Liability

TL;DR: Asked when the majority will be using AI, most owners answer "a few years." The data says the majority already is — through the AI embedded in their search results, social feeds, office software, cars, phones, and operating systems. Deliberate AI use crosses 50% of U.S. adults around 2027 on Pew's trajectory. Meanwhile the old playbook — WordPress websites, keyword content, voicemail — just became a security liability under a federal patch-or-offline warning, and an economy shifting from two-income households to one is making efficiency the KPI that decides which contractors survive. The window to get positioned is measured in months, and in the channels your customers actually use, it has already closed.

"How much time do we have?"

It's the question every owner in the trades is asking right now. Time to get ready. Time before AI changes how customers search, compare, and choose. Most people assume the answer is years — that AI is a future project, something to budget for next year, pilot someday, and watch from a distance while the early adopters experiment.

The data says otherwise. The window is measured in months — and for the channels your customers actually use, it has already closed. Here's why.

The majority isn't coming. It's here.

Count AI the way people actually encounter it — inside the apps, search results, cars, phones, and computer operating systems they already use — and majority adoption isn't a forecast. It's the present.

• 2 billion+ monthly users now see Google's AI Overviews — AI answers sitting on top of the world's largest search engine, reported July 2025, across more than 200 countries.

• 1 billion monthly users use Meta AI inside Facebook, Instagram, WhatsApp, and Messenger — announced May 2025, double the count from just eight months earlier.

• 60% of U.S. adults already use AI to search for information, according to AP-NORC polling from July 2025 — including 74% of adults under 30.

A majority, today, by self-report — before we count a single person who "hasn't tried AI yet" but reads an AI answer at the top of every Google search.

The fastest curve in history

Every technology climbs the same S-curve. AI is climbing it faster than any before it. Approximate years to reach 50% of U.S. adults or households:

• Telephone: ~45 years

• Television: ~24 years

• Internet: ~8 years

• Social media: ~6 years

• Smartphone: ~5 years

• AI agents: ~4 years (projected)

ChatGPT reached 100 million users in two months — the fastest consumer adoption ever recorded — and reported roughly 900 million weekly users by February 2026. Pew Research's independent surveys tell the same story from the U.S. side: adults who have ever used ChatGPT went from 18% in 2023 to 34% in 2025 to 44% in 2026. At that pace, deliberate AI use crosses the 50% majority line around 2027.

And remember: that's the deliberate number — people who know they're using AI. The embedded number crossed the majority long ago.

The evolution: chat → agents → networks

One more thing the adoption numbers alone don't show: what is being adopted is itself evolving, in three stages. They don't replace each other — they stack.

Stage 1 — Conversational chat. You ask, it answers. One question, one response, a human driving every step. This is the AI most people picture, and the one the chatbot growth numbers measure. It's a tool — powerful, but passive. Nothing happens unless a person prompts it.

Stage 2 — AI agents. You state a goal, and the AI works the steps itself: researching, comparing, using tools and applications, drafting, booking, buying, and reporting back. The human moves from operator to director — setting the direction and approving the consequential steps. This stage is deploying right now: McKinsey found 62% of organizations experimenting with AI agents and 23% already scaling them. The homeowner version: instead of just answering "which AC should I buy," the agent assembles the options, checks availability, and books the install.

Stage 3 — Fully autonomous agentic networks. Agents working with agents. A contractor's agent negotiating availability and price with a distributor's agent. A homeowner's agent gathering quotes from three contractors' agents and scheduling the winner — with no human inside any of those loops. People still set the goals, the budgets, and the guardrails; the network executes, and humans step in by exception. This stage is forming now, in pieces, on top of the agents already deployed.

Here's why the staging matters: the decisions made in stage 2 determine who holds power in stage 3. When the networks form, they will route work to the businesses agents can find, verify, and transact with — the same citation-and-verification game, played machine-to-machine. Chat is mainstream. Agents are scaling. Networks are forming. Planning for stage 1 while your market enters stage 3 is how a business gets disintermediated quietly.

Five channels, one conclusion: adoption is a default now

And be clear about what's actually arriving: it isn't chatbots. It's AI agents — autonomous systems that don't just answer questions but act on them: researching, comparing, booking, buying, and working in the background whether anyone is watching or not. Previous technologies waited for people to choose them. This one is pre-installed. Look at the five channels carrying it:

1. The Feed. Meta didn't launch an AI product people had to find — it moved AI into Facebook, Instagram, WhatsApp, and Messenger, apps people already open dozens of times a day. There was no download, no signup, no behavior change. AI appeared in the search bar and the chat list. Asking Meta AI is now easier than opening Google.

2. The Office. Roughly 70% of the Fortune 500 run Microsoft 365 Copilot, and Microsoft reported 15 million paid Copilot seats by January 2026 — up 160% year over year. Gallup found 45% of employees already use AI at work, 23% of them weekly and 10% daily. Why it matters to you: behavior learned at work comes home. A homeowner who spends all day asking Copilot instead of searching will ask AI — not Google — when the AC dies at 9 p.m.

3. The Driveway. Tesla had delivered 8.9 million vehicles by the end of 2025, essentially all equipped with Autopilot as standard. Their fleet had driven roughly 7 billion miles on Full Self-Driving, and 1.1 million owners pay monthly for FSD subscriptions — up 38% year over year. Driving is the most trust-intensive task there is. When someone trusts AI at highway speed, twice a day, trusting it to recommend a contractor is a small step.

4. The Pocket. Apple's installed base hit 2.5 billion active devices in January 2026 — the largest hardware footprint on earth. Siri has handled billions of voice requests a month for years; Apple Intelligence now layers generative AI on top. Today, more than half of iPhones can't run Apple Intelligence — and that gap is not a ceiling, it's a queue. Every upgrade cycle converts millions more people to an AI-first phone, on Apple's schedule, without them ever making an "AI decision."

5. The Desktop. AI is now embedded in the computer's operating system itself. Copilot ships inside Windows across 1.4 billion+ monthly active devices. Apple Intelligence is built into macOS. Gemini is built into Chromebooks. No download, no decision — the OS brings AI with it, at home and at work.

Five channels, one conclusion: adoption isn't a decision anymore. It's a default.

The map: the gap is months, not decades

Adoption runs hottest in major metros and among younger buyers — AP-NORC's 74% under-30 figure versus 60% overall tells that story. If your contractors serve a top-50 metro, their AI moment is now.

But don't take comfort if you're in the center of the country. Broadband took decades to cross America because it needed wires. AI needs none — the same Meta AI, Apple Intelligence, and AI Overviews ship to every market on the same day. The lag that remains is habit and device age, and both shrink every quarter.

The distance between Austin and Amarillo used to be measured in infrastructure. Now it's measured in upgrade cycles.

The next shift: from indexed, to cited, to verified

There's a second shift riding on the first, and it's the one that decides whose phone rings.

The first generation of AI was trained on the open web — your content was in the pile whether you optimized or not. That era is ending. AI companies increasingly ground their answers in licensed, proprietary, and verified sources they can check in real time. Visibility is moving through three stages:

• Then — be indexed. Get crawled, rank for keywords, win the click. Visibility was a position on a results page.

• Now — be cited. AI reads, compares, and answers — naming the businesses it trusts. Visibility is being

  in the answer.

• Next — be verified. Consistent facts, reviews, credentials, and structured data AI can confirm — or you simply don't exist in its answers.

The behavior data shows what that means in practice. Pew's analysis of real Google searches found AI summaries appearing in about 18% of searches — and when one appeared, only about 8% of searches led to a click on a traditional result, while roughly two-thirds of all Google searches ended with no click at all. The answer is increasingly the destination. Being cited inside it is the new being found.

Your website's new job

Here's the part that gets misread most: your website is not being discredited. Its job changed.

It used to be the destination people clicked to at the end of a search. Now it is the source AI reads, quotes, verifies, and cites — before it ever sends a customer your way. Which means the entire content strategy on the website has to evolve:

• The old strategy: thin keyword blogs written for algorithms. Service pages stuffed with search terms. Content produced to rank, not to answer. AI can tell the difference — and it cites accordingly.

• The evolved strategy: real answers to real customer questions. Real proof — job data, before-and-afters, pricing context. Reviews, front and center. Video with real transcripts. And structured, verifiable facts — licenses, service areas, credentials — that AI can check and safely cite.

Invest heavily here. Your website is no longer the brochure at the end of the search. It's the asset the answer is built from.

The old playbook isn't just outdated — it's exposed

Name the framework most of the industry is still running: a WordPress website built years ago, thin blog posts written for keywords, attention rented from ads, and a phone that rolls to voicemail after hours. The old playbook assumed a customer who searches, clicks, and calls during business hours. Every one of those assumptions is now broken — the customer asks an AI, the AI cites its sources, and the job goes to whoever the answer names.

And this fall, the old stack stopped being merely outdated and became a security emergency. In October 2026, the NSA, FBI, and CISA issued a warning to every organization running PHP — the programming language underneath WordPress: patch immediately, or take your systems offline. Two remote-code-execution flaws, graded 9.2 out of 10 in severity, had been hiding in PHP for roughly 14 years — and PHP versions 8.1 and lower have no patch at all.

Why this lands on distributors: by industry estimates, 70–90% of contractors are running exactly that stack — PHP-based WordPress websites carrying the old content strategy. A hacked or offline site cannot be cited, ranked, or trusted; it doesn't just lose its own leads, it disappears from the AI answers its market runs on. And distributors and manufacturers will feel this upstream: your sell-through depends on dealer networks that consumers can find. A network whose websites go dark — or get flagged as compromised — stops generating demand for everything behind it. Security is no longer IT hygiene. It is part of AI visibility, and it is now a channel problem.

The asset math: owned and earned beat paid

In the AI era, the three kinds of media are no longer equal:

• Paid is rented. The day the budget ends, the visibility ends with it. And you cannot buy a citation at any price — AI doesn't sell its answers. An ad has never made AI recommend anyone.

• Owned compounds. Every real answer you publish keeps working — cited for years, not until the budget runs out. Your website is where AI verifies who you are and what you do, and no algorithm change or auction can take it away.

• Earned is the trust signal. News coverage, reviews, YouTube — third-party proof AI leans on when it decides whom to name. It can't be faked at scale, which is exactly why the models weight it.

The data backs the ranking: Meltwater's GenAI Lens study, tracking roughly 5.35 million citations across eight AI models, found 39.5% of AI citations come from earned and news sources — and YouTube is a top-five citation source for six of the eight models.

And the timing is critical. Citation positions are forming right now, while most competitors do nothing. Authority takes months to build and cannot be bought at the moment of need. When deliberate AI use crosses 50% around 2027, the cheap window closes — the businesses cited then will be the ones who built while it was still cheap to own the answer.

The lead math: AI buyers arrive decided

One more set of numbers, because it changes how you should judge every marketing report you see. A lead from AI search is a different animal from a lead from ads or organic search — they've already had the sales conversation, with the AI:

• Seer Interactive's multi-client study measured AI-search leads (Perplexity) converting at 15.9%, versus 9.77% for Google organic.

• Semrush values an AI-search visitor at 4.4× an organic visitor.

• In Ahrefs' own data, AI search sent just 0.5% of traffic but 12.1% of signups.

Smaller volume, radically better math. These buyers arrive pre-sold; the call is about scheduling, not convincing.

So why do most reports still show AI as tiny? Because the measuring instrument is broken. When someone clicks through from ChatGPT or another AI app, the referrer is often stripped — and analytics files that visit under "Direct." Until May 2026, the industry's standard analytics had no AI channel at all, and the one added isn't retroactive. Every AI number before that is a floor, not a total. The leads were there. The instrument couldn't see them.

Which produces the trap: benchmarks built on that data understate AI, owners conclude "AI doesn't produce," and keep funding the old mix — a decision made on a broken ruler. The data reflects the strategy that produced it, not the market that's actually out there. You can't generate what you don't market for.

The squeeze: efficiency is the new KPI

There's a second force compounding the first, and it's economic.

AI is displacing corporate jobs at volume. As that displacement spreads, household math changes with it: two-income households become one-income households, and the customer on the other side of your contractors' estimates has less money, less cushion, and a far sharper eye on price. At the same time, equipment and materials costs keep rising. Your contractors are being squeezed from both directions at once — customers who can spend less, and jobs that cost more to deliver.

In that market, the winners are already visible: tiny teams, fully AI-enabled. Contractors who run lean — agents handling the answering, scheduling, follow-up, quoting support, and content — carry a cost structure the old model simply cannot match. That lets them compete on price without surrendering the margin. Efficiency stops being an operations metric and becomes the KPI — the number that decides who survives the squeeze.

The consumer side of the new model is just as unforgiving:

• Pricing transparency, published. Prices and honest price ranges live on the website, in the open — because that's where AI reads them, and where customers now expect to find them. Hidden pricing reads as a red flag to both.

• Financing, no exception. In a one-income market, financing isn't an upsell — it's how the job gets approved at all. Contractors who lead with financing close work that "call us for a quote" contractors never even hear about.

For distributors, the move is urgent and familiar: train your contractors to be lean operators now — while the shift is still an advantage to seize rather than a standard to catch up to. The distributor whose network runs efficient, AI-enabled operations owns the price-competitive tier of its market. Don't wait.

Google's agents are calling your contractors — and asking for prices

One more change, already live: the caller asking your contractors for pricing may not be a person.

Google's AI now calls businesses on consumers' behalf — asking for prices and availability, comparing the answers across competitors, and reporting back to the customer. Those calls are recorded, and what your team says is on the record: information given to Google's automated calls can end up reflected on the business's Google Business Profile. A guessed price on the phone can become a contractor's published price.

The new rules of the phone follow from that machinery:

• AI answering is allowed. Google has published no rule requiring a human voice. What matters is whether the call gets handled — real answers, real booking.

• Speed is the standard. A voice that can help within seconds. Since October 1, 2026, a missed Local Services call during posted business hours is a charged lead if the caller waits more than 20 seconds. Ringing out to voicemail is now a paid failure.

• Whoever answers must be able to book — name, address, problem, a real appointment window, on that same call. A message-taker is a failed call, human or AI.

• Quote the published price. Google's agent asks the same questions a customer does, and reports non-answers just as faithfully as answers. The rough price quoted on the phone has to match the price on the website — one story, everywhere.

• Never hang up on the agent, and never route it to voicemail. Treat Google's AI call as a customer interview — because it is one, with the summary delivered straight to the customer.

We built a complete Owner's Guide to these rules — Answer: The New Rules of the Phone — because owners kept asking what the rules even are. The short version for your network: the phone is no longer a courtesy channel. It's an audited one.

What this means for distributors, specifically

Translate all of this to your seat in the chain, and three things stand out:

1. Your contractors' visibility is your volume. When a consumer's assistant names your contractor, that contractor buys equipment — from you. The distributor's growth now runs through AI answers it doesn't directly control, which is exactly why it has to lead on them.

2. You can move a whole network at once. One distributor teaching its contractor base what "cited" means — and helping them build the content that earns citation — shifts an entire market's answers. No single contractor has that leverage. You do.

3. Your own citability counts. Your product knowledge, application expertise, and brand authority are content AI reads too. And with 88% of organizations now using AI in at least one business function (McKinsey, State of AI 2025), your manufacturer and vendor partners are moving as well — the network advantage goes to the distributor who connects it all first.

4. Your network's stack is your exposure. With an estimated 70–90% of contractors on PHP-based WordPress and a federal patch-or-offline warning in force, your dealers' websites are now a supply-chain risk. The distributor who audits the network's foundation — and helps fix it — is protecting its own volume.

Your moves this quarter

1. Reset your timeline. Stop planning to "years." Embedded AI is already the majority; deliberate use crosses 50% around 2027. Plan to months.

2. Run the citation audit. Ask the major AI assistants the top ten questions your contractors' customers ask. Whose names come up? If it's not your contractors — and not you — that's your gap, and it's measurable.

3. Rebuild content strategy around citation. Real answers, real job proof, reviews up front, video with transcripts, verifiable facts. Fund owned and earned first; keep paid in its supporting role.

4. Fix the measurement before you judge it. Ask whoever sends your reports how AI-referred traffic is counted. If the answer is "it shows up as Direct," you're making decisions on a floor, not a total.

5. Train your contractors. Teach them what "cited" means, why it matters, and what content earns it. The distributor who trains the network owns the network's future.

6. Publish the money page. Pricing ranges and financing options on the website, in the open — matching exactly what the phones quote. Transparency is now table stakes, for customers and for agents.

7. Pass the phone test. Mystery-call your contractors the way Google's agent does: answered in seconds, a real price quoted, the job booked on the call. Anything less fails the new rules of the phone.

8. Audit the stack. Find out how many of your contractors sit on unpatched PHP/WordPress, and treat the fix as a visibility project, not an IT ticket — a dark site is a dark market.

The bottom line for HVACR Distributors

When will the majority be using AI? Embedded AI: already the majority. Deliberate AI: crossing 50% around 2027. The only timeline that matters now is yours.

The majority isn't coming. It's here. Be the answer.