Two Businesses, One Phone Line, Zero System Built for Either

Before anyone builds anything, you need to know what problem you're actually solving and whether it's worth solving. Here's how to think through it with your own data.

Start with after-hours calls. Pull your phone records for the last 90 days and count how many calls came in outside business hours. Of those, how many resulted in a booked job? Now ask yourself honestly: what happened to the ones that didn't? A missed emergency call isn't a lost $150 service fee — it's a lost spring replacement, possibly a new opener, and almost certainly a lost customer for any future work. A single new door install lost to a competitor because you didn't answer at 8 PM on a Saturday represents real margin walking out the door.

Now look at your maintenance side. How many customers have had service in the past 18 months but haven't been contacted about their next visit? If you offer maintenance agreements, what's your renewal rate, and do you actually know when agreements lapse? According to the International Door Association, maintenance agreements and recurring service contracts represent a significant portion of revenue for established garage door dealers — but that revenue only materializes if someone is tracking it.

The dispatcher question is harder but equally important. How many hours per week does your dispatcher spend on tasks that are purely logistical — confirming appointments, calling customers with ETAs, fielding status questions? If that number is more than a few hours per week, you're paying a skilled coordinator to do work that doesn't require human judgment.

• What would one additional captured emergency call per week be worth annually? Multiply your average ticket by 50.
• What's your current maintenance renewal rate? If you don't know it, that's the first problem.
• How much time does dispatch spend on routine confirmations vs. actual problem-solving? Track it for one week.

You don't need a consultant to tell you the magnitude — your own call log and revenue data will do it. The goal is to build a system before you've already lost the jobs, not after.

AI doesn't work in isolation. It needs to read from and write to the systems your business already runs on — and in garage door service, that stack is usually a mix of field service software, a phone system, and a calendar that may or may not be connected to anything else.

The most common scheduling platforms in this space are ServiceTitan, Jobber, and HouseCall Pro. Each has API access, but the depth of that access varies. ServiceTitan's integration layer is mature but requires careful configuration. Jobber and HouseCall Pro are more straightforward for smaller operations. If you're running off QuickBooks and a shared calendar, the integration work is heavier — not impossible, but plan for it.

Your phone system matters more than most owners expect. An AI intake agent needs to connect to your existing business line or be provisioned as an overflow route. If you're on a VoIP system like RingCentral, Dialpad, or even a basic Twilio-based setup, that's workable. If you're running calls through a personal cell phone that also handles business, you need to fix that before any AI layer makes sense.

(Source: CompTIA, 2023 — small service businesses that lack documented workflows in their field service software see significantly longer automation deployment timelines than those with clean data and consistent job categorization.)

• Document your job types before integration starts. Emergency service, scheduled maintenance, new installs, and warranty callbacks need to be categorized consistently — not just in someone's memory.
• Clean your customer records. Duplicate contacts, missing phone numbers, and service history gaps will break automated outreach before it starts.
• Know where your parts data lives. If techs are confirming parts availability by calling the supplier from the job site, that's a gap that automation can address — but only if inventory data exists somewhere.
• Establish an escalation path. AI handles routine intake well. It needs a defined handoff for situations it can't resolve — angry customers, liability questions, unusual job types.

Oaken's stack — FastAPI, PostgreSQL, and direct integrations with ServiceTitan and Jobber via their published APIs — handles the plumbing. But we can only build on top of data that exists and is structured. The owners who move fastest are the ones who've already standardized their job categories and keep service history in their scheduling software, not on paper job tickets in a filing cabinet.

The garage door industry is not the first stop for most AI vendors, which means the pitches you're hearing are usually adapted from adjacent trades — HVAC, plumbing, general home services. That's not automatically a problem, but it means you need to ask sharp questions about what's actually been built for your workflows versus what's been rebranded.

The most common oversell right now is the "fully automated dispatcher" claim. Any vendor telling you their system will replace your dispatcher entirely is either misinformed or hoping you don't ask follow-up questions. Emergency service dispatch involves judgment calls — a tech who's already on an emergency job can't take a second one, a customer in a dangerous situation needs a different response than someone mildly inconvenienced, a job that sounds like a broken spring on the phone might be a failed motor. AI handles intake and routing logic well. It does not handle edge cases well without a human in the loop.

Watch for vendors who demo their product against ideal scenarios — clear audio, cooperative customers, standard job types. Ask them what happens when a customer can't describe the problem, when the job type is ambiguous, or when the tech calls in sick mid-route. The answer to those questions tells you more than any demo.

• "Guaranteed ROI" claims without seeing your data — no one can tell you what you'll capture until they know your current call volume, average ticket, and close rate.
• Setup fees that don't include integration work — connecting to ServiceTitan or Jobber correctly takes real engineering time. If the quote doesn't account for it, you'll pay for it later.
• One-size workflows for emergency and maintenance calls — these are different conversations that need different logic. A vendor treating them the same hasn't worked in field service.
• Review-generation tools positioned as the primary AI use case — review automation is real and useful, but if that's the centerpiece pitch, they're not solving your actual operational problems.

The vendors worth talking to are the ones who ask about your dispatcher's current workload before they pitch anything, who want to see your job categories and call log before scoping a project, and who can explain specifically how their system handles a call it can't resolve on its own. Caution is warranted — but so is moving. The shops that figure this out first have a real competitive advantage in markets where emergency response time is the primary differentiator.

Not every garage door company should be building AI systems right now. That's an honest statement, and it matters more than any pitch about what's possible.

The operations that are genuinely ready share a few consistent characteristics. They have at least two full-time field technicians and a defined dispatcher role — even if that role is part-time or shared. They're already using scheduling software, even imperfectly. They have some record of service history per customer. And they're experiencing a specific, observable problem: calls going unanswered, maintenance renewals falling off, dispatcher overwhelmed during busy periods. The Bureau of Labor Statistics projects employment in installation and repair occupations to grow steadily through 2032, which means the labor market for dispatchers and coordinators isn't getting easier — building systems that extend your existing team is the practical path. (Source: U.S. Bureau of Labor Statistics, 2023)

The operations that aren't ready yet are equally identifiable. If your service history lives in a filing cabinet or in a dispatcher's notebook, you have a data problem that precedes any AI solution. If your job categories change depending on who takes the call, you don't have consistent enough workflows to automate. If you have one technician and you're that technician, the bottleneck is capacity, not coordination — and AI doesn't fix a staffing shortage.

• Ready: 2+ techs, digital scheduling, defined job types, dispatcher spending significant time on routine coordination
• Ready: Operations with maintenance agreements that aren't being tracked or renewed consistently
• Ready: Companies losing identifiable after-hours calls to competitors
• Not yet: Owner-operator with no admin support and no digital records
• Not yet: Operations where every job is a custom situation with no repeatable workflow
• Not yet: Businesses that haven't decided what their service area is or what job types they take

The honest prerequisite isn't size — it's process maturity. A four-tech shop running clean data out of Jobber is more ready than a ten-tech operation where every dispatcher has their own system. If you're not sure where you fall, the clearest test is this: could a new dispatcher learn your intake and scheduling process from documentation alone, without asking you questions? If the answer is no, start there before you start with AI.