Leads generated by my AI systems
485 new leads a day by October, 4.3x the July pace. Every one captured, followed up and traced to its source by AI.
For plumbing companies · free 30-minute strategy call
I'm Jon Dykstra. I run my own businesses on AI: millions of visitors, tens of thousands of leads, and 1,878 ad campaigns launched by AI, every one judged every single morning. Every lead gets captured, followed up and traced to its source.
Plumbing companies lose jobs in the same places my businesses used to: calls nobody answers, quotes nobody chases, and ads judged on clicks instead of booked work. That's exactly what this machine fixes.
Leads = opted-in email subscribers and home-equity lender leads across businesses Jon owns, Jan 1–Oct 14 2026. Not plumbing client results.
Free, 30 minutes, phone or Google Meet. No pitch deck.

Where plumbing companies lose jobs
If you've been in business a while, the phone already rings. Your trucks are known in town, your reviews are decent and you're spending real money on Google and maybe Facebook. The money that goes missing goes missing after the phone rings, somewhere between the first call and the booked job.
Here's where I'd look first in almost any established plumbing company:
None of those are marketing problems in the usual sense. They're follow-through problems. Follow-through is the thing people are worst at when they're busy, and it's the thing AI does best.
The proof
This isn't a course or a slide deck. It's the machine I run on my own businesses right now. In a single week in October, AI launched 299 ad campaigns, published 145 pages and brought in roughly 485 new leads a day, every one followed up and traced to its source.
Jon's own portfolio. Figures are from businesses Jon Dykstra owns and operates on AI. They are not client results. Data to Oct 11 2026 unless noted. Leads = opted-in email subscribers and home-equity lender leads across businesses Jon owns, Jan 1–Oct 14 2026. Not plumbing client results.
Leads generated by my AI systems
485 new leads a day by October, 4.3x the July pace. Every one captured, followed up and traced to its source by AI.
A plumbing lead is a phone call about a burst pipe at 9 pm. My leads are email sign-ups and lender leads. The machine underneath is the same.
It captures every lead, responds in minutes, follows up until the job is booked, traces every booking to its source, and moves money toward what works every morning. I run that machine every day on my own businesses, with my own money: 46,800+ leads so far this year. I've watched it fail, fixed it and kept what worked. That's what I build for plumbers.
You don't need 46,800 leads. You need the calls you're already paying for to turn into booked work, and you need to know which ads actually fill the schedule.
The AI operating system for a plumbing company
Most plumbing companies have the pieces already: a phone system, Google Ads and Local Services Ads, maybe Facebook, a website, a Google Business Profile, and ServiceTitan, Jobber or Housecall Pro running the schedule. The pieces just don't talk to each other, so nobody can see the whole path from an ad to an invoice.
We connect them to one core: the rules that decide what to do, and a memory of what booked jobs. You set the budget, the limits and the words AI is allowed to use with your customers. AI answers, follows up, books, measures and tunes inside them, every day.
Workflow 1 · getting the phone to ring
Every source gets its own tracking number and its own tagged links, so a call from a Local Services Ad, a call from a Facebook ad and a call from your Google Business Profile all arrive labelled. Every call and form is logged with its source, the type of job and the area it came from. Then it's sorted: emergency, quote, or maintenance.
On the content side, AI builds and keeps up the pages that pull in local searches: a page for each service and each town you want more work in, plus regular Google Business Profile posts and before-and-after photos from your jobs. In my own businesses this is the department that went from 35 pages a week to 145, so volume won't be the limit for you.
On the paid side, AI writes the ads, launches them inside your caps and builds new tests every week. Seasonal campaigns get planned ahead of time instead of the night before.
Workflow 2 · the first reply
Think about how a homeowner hires a plumber. Water on the floor, a phone in one hand, three names from a search. They call or fill in a form, and the first company that answers with something useful usually wins. Most companies lose that race on paperwork: a form that lands in an inbox, a web chat nobody watches at 6 pm, a Facebook message read the next morning.
With AI on it, every form, chat and message gets a reply in minutes, day or night, in words you've approved. The reply asks the questions your dispatcher would ask: what's happening, where is it, how urgent, and when can someone be there. Emergencies go straight to a person. Routine work gets offered a time.
A homeowner fills in your website form at 6:40 pm: "Kitchen sink backing up, second time this month." By 6:42 she has a text: "Thanks for reaching out. Sorry about the sink. Is water coming up into the sink now, or is it draining slowly? We have a tech available tomorrow at 8 am or 11 am. Which works?" She picks 8 am. The job is in the schedule before your office reopens.
Workflow 3 · missed-call rescue
Your office can't answer every call. Nobody's can. Techs are on jobs, the line is busy at 8 am, and after hours goes to voicemail. Most callers don't leave a message. They call the next name.
This workflow catches them. When a call goes unanswered, the caller gets a text within about a minute: "Sorry we missed you. What's going on?" AI reads the reply and sorts it by your rules. A flooded basement at 9 pm alerts your on-call tech. A dripping faucet gets a booking link for tomorrow. Every one is logged, so no lead disappears and you can see exactly how many calls you'd have lost.
You write the rules. You decide what counts as an emergency, who gets alerted, which hours and areas you cover, and what AI may say. That's the gate in the diagram.
Workflow 4 · booking
Once a homeowner says yes, the booking goes into the software you already run the company on: ServiceTitan, Jobber or Housecall Pro, wherever the tool's API and your plan allow it. The customer gets a confirmation, a reminder the day before, and a heads-up when the tech is on the way if your software supports it. If they need to move the appointment, they can reply to the text instead of calling.
Your office stays in charge of the board. AI fills the requests in and keeps customers informed; your dispatcher still decides who goes where.
Workflow 5 · quotes that close
Water heaters, repipes and sewer lines are the jobs that make your month. They're also the jobs homeowners think about, compare and put off. A quote goes out, the homeowner says "let me talk to my husband", and then nothing happens because your office is busy with today's calls.
AI follows up on every open quote on a schedule you set. A typical starting cadence is day 2, day 5 and day 10, each message written for that job type and that customer. It answers simple questions, offers to book the work, and hands anything tricky (price, scope, a second opinion) to your estimator. When a quote is won or lost, the reason is recorded, so the follow-ups improve.
Workflow 6 · the customers you already have
Every plumbing company has a database full of people who already trust it. Most of them never hear from the company again until something breaks, and by then they may have forgotten your name.
AI works that list on a calendar, using your job history:
Every message is in your voice, uses words you've approved, and respects opt-outs. Replies come back into the same booking flow as any new lead.
Workflow 7 · daily performance reporting
Overnight, AI collects yesterday's ad spend, calls, forms, bookings and jobs from every system you use, and checks the totals against their sources. At 7 am you get a short text or email: how many calls came in, how many were missed and recovered, how many jobs were booked, what each ad source cost per booked job, which quotes need a call from you, and what AI already did about yesterday.
If the numbers don't reconcile, the report says so instead of guessing. I built that rule into my own reporting because a confident wrong number is worse than no number. Accounting connections are read-only. AI reports on the books; it doesn't move money.
Workflow 8 · continuous campaign improvement
This is the part I know best. In my own businesses AI launched 1,878 campaigns between June and October, and every one of them is judged every morning on what it actually returned. Losers are paused by rule. Winners get more budget. That's how weekly Facebook spend under AI's management grew from US$460 to US$19.8K without me babysitting a single bid.
For a plumbing company, "what it actually returned" means booked jobs. Every call and form is matched to the ad behind it, and every booked job is matched to the call. Each morning AI looks at cost per booked job by campaign, by service and by area, and moves money inside your caps. It never raises your total budget without your approval.
Jon's own portfolio. Figures are from businesses Jon Dykstra owns and operates on AI. They are not client results. Data to Oct 11 2026 unless noted. Campaign and ad-spend figures: AI launch queue, Jun 22–Oct 11 2026, and Meta Ads API, Feb 16–Oct 11 2026.
Most marketing stays the same until somebody has time to fix it. This system is built to improve every day, through a loop you can see and control:
Which service areas pay. Which offers get booked. How fast to follow up and what to say. Week 20 is far sharper than week 1, and nothing has to be relearned.
One real example from my own businesses: my early ad rules judged new campaigns too soon and turned out to be killing future winners. The system tested its own rule against results, learned to wait for more data, and profitable campaigns that would have been switched off stayed alive. In your company, the same discipline applies to rules about when to pause an ad or when to send a follow-up.
Connections to ServiceTitan, Jobber, Housecall Pro and other software depend on each tool's API and your plan, and are confirmed during the audit.
What it looks like
The owner's part is in gold. Everything else runs on its own.
AI judges yesterday's ads on booked jobs, not clicks. Two Google campaigns that booked no work are paused. The drain-cleaning ad that booked four jobs gets more budget, inside the daily cap.
The owner gets a text: 23 calls yesterday, 3 missed and all 3 texted back within a minute, 14 jobs booked, cost per booked job by source, and one water-heater quote that needs a personal call.
YouA homeowner asks about a water-heater replacement through the website. AI answers, books an estimate into the schedule and sends a confirmation.
Three water-heater quotes from last week get their day-5 follow-up text. One replies "yes, book it."
A sewer-line customer replies to a follow-up with a question about the warranty. AI passes it to the estimator with the quote and camera notes attached.
AI publishes a new service-area page for a nearby suburb and posts a before-and-after on Facebook and the Google Business Profile.
Customers whose water heater was installed a year ago get a flush reminder with a self-booking link. Two book for next week.
AI proposes a test: a "same-day drain clearing" offer for two zip codes where calls are slow. The owner approves it from the phone.
YouA burst-pipe call goes unanswered. Within a minute the homeowner gets a text. AI flags it as an emergency and alerts the on-call tech.
Every call, job and dollar is reviewed. The follow-up timing and ad rules are updated from what actually booked.
The owner's part of that day is about ten minutes: reading one report, making one call and approving one test.
Your command center
This is what running a plumbing company on AI looks like from the owner's chair. Today's calls, with missed calls already texted back. The follow-up queue and every open estimate. The dispatch board and jobs booked. Cost per booked job by ad source. Maintenance reminders going out, the 7 am report, and an assistant you can ask "what did Google cost us per booked job last month?"
Click around. Approve a recommendation, open the estimate pipeline, ask the assistant a question. It's a fictional company with sample data, and the workflow is the real one.
Interactive demo with fictional sample data. Every action is simulated.
Sample data · click to explore
Optional add-on · custom AI tools
Most plumbing websites say the same things: years in business, a list of services, a phone number. A tool answers the question the homeowner actually came with, and then offers an exact quote. It's optional, and most clients start with follow-up, ads and tracking. We can build whatever fits your company. A few ideas:
Leads from a tool flow into the same instant follow-up, booking and tracking as every other lead, so a homeowner who sizes a water heater at 9 pm gets a reply in minutes.
Who you're talking to

In 2025 I went through my businesses one job at a time and gave every job AI could do better, faster and cheaper to AI: the content, the ads, the bidding, the social, the email, the lead follow-up and the reporting. Plenty broke. I fixed it with my own money on the line, and kept what worked.
The result is a lead machine running at real scale: 46,800+ leads, 27.3M pageviews and 1,878 AI-launched ad campaigns this year, with every campaign judged every morning. As I put it in my report, "I get roughly 25x done in a quarter of the time." That's my own estimate, and it's the reason I started building this for other owners.
I trained as a lawyer and earned an MBA, so I judge every system like an owner: does it make money, does it save time, and can I trust the numbers? I'm not a plumber, and I won't pretend to know your trade better than you do. What I know is how to capture, follow up and measure leads at real volume, every day, and plumbing is a business where that discipline pays.
A client build for a plumbing company is underway now, and I'm currently accepting a few more clients. You get the system without paying for my expensive year of lessons.
Jon's own portfolio. Figures are from businesses Jon Dykstra owns and operates on AI. They are not client results. Data to Oct 11 2026 unless noted. Leads = opted-in email subscribers and home-equity lender leads across businesses Jon owns, Jan 1–Oct 14 2026. Not plumbing client results. 190x compares the week of Oct 5–11 2026 with the average January week; 91% of the final week came from one viral page on a newer site.
You stay in charge
You set the goals, budgets and limits. AI does the work inside them. Anything that spends money, contacts customers or changes the business can require your approval, and everything is logged.
And it's yours. Everything is built on your own accounts: your ad accounts, your phone numbers, your website, your software, your data. If we ever part ways, it all stays with you.
How it works
Everything is built on your own accounts. You own all of it. We run it for you, or train your office to run it.
We map your marketing and operations and rank every job AI could take over by the money and hours it frees up. In most plumbing companies, tracking and missed-call follow-up sit near the top.
The AI systems, tracking and dashboards go in on your own accounts.
Want your team to run it? We train them on every system until they're confident without us.
Rather not? We run it for a monthly fee. AI does the work; we watch the numbers and tune the system.
AI brings new tests built on your results. You approve them and AI carries them out.
Because the hard part was never the trade. The hard part is answering every lead fast, following up until it books, tying every dollar to the result it produced and doing that every day without getting tired. I do that across millions of visitors and tens of thousands of leads a year, in businesses I own. Your trade knowledge plus that machine is the combination I'm offering.
Questions
No, and I'll never pretend they are. The 46,800+ leads and the charts on this page come from businesses I own and run on AI. They're email sign-ups and home-equity lender leads, not service calls. A client build for a plumbing company is underway now. What I'm offering is the same machine, built for your company, on your accounts.
We connect to the field-service software you already use wherever its API and your plan allow it, so calls, bookings and jobs flow into the reporting automatically. If a connection isn't possible, I'll tell you on the call and we'll work around it.
Only where you want it to, in words you approve: missed-call texts, booking confirmations, quote follow-ups and maintenance reminders. Anything unusual, upset or complicated goes to a person.
No. You set daily and monthly caps. AI moves budget between campaigns inside those caps and pauses losers by rule. It never raises the total without your approval, and every change is logged.
No. Your office keeps the phones and the dispatch board; AI takes the follow-up and the busywork they never get to. If you have an agency, we'll look at what it does on the call. You might keep yours for some channels and let AI handle tracking, follow-up and reporting.
Tracking and missed-call follow-up usually go live in the first weeks of a build, because they pay back fastest. A full build takes about a month.
You do. The ad accounts, phone numbers, website, software, data and every rule the system learns are on your accounts. Accounting access is read-only.
It depends on what AI should take over first and whether we run it or teach your team. You'll get a straight answer and a clear scope on the free call. If you'd like to see how I think first, read the free report. Read the free report first →
Free strategy call
Phone or Google Meet, your choice. Times show in your time zone. Only a small number of businesses are taken on at a time. Not ready? Read the free report or send a message.