Every client engagement we run starts the same way: not with a build, but with a roadmap. It's the setup fee on our pricing page, and it's the least glamorous, highest-leverage thing we sell. This post explains exactly what happens during those two to three weeks, what you walk away with, what it costs, and — honestly — who shouldn't buy one.
Published June 9, 2026 · 10 min read
An AI roadmap is a written document that answers four questions about your specific business: where your operation leaks time and money, what data and tools you actually have to work with, which AI opportunities are worth building ranked by return, and in what order to build them. That's it. No futurism, no slideware about the coming age of agents. You can sketch a rough version of that ranked plan yourself in two minutes with our free AI automation opportunity finder — and gut-check whether you're set up to act on it with the AI readiness scorecard.
The reason it exists is blunt: most failed AI projects in small business fail before the first line of code — they automate the wrong thing. A beautiful voice agent for a business whose real leak is unworked estimates. A dashboard for an owner whose QuickBooks hasn't been reconciled since March. The build was fine. The diagnosis never happened.
So we made the diagnosis mandatory. Every retainer we sign starts with this engagement — it's what the setup fee on our pricing page pays for ($1,500–$3,500 depending on plan). We won't skip it even if you arrive certain about what you want, because clients arrive certain and are wrong about the first build roughly half the time. Usually not about whether the idea is good — about whether it's the best first use of a build slot.
What it asks of you is modest but real: four to six hours of your team's time spread across two to three weeks — an interview with whoever runs your office, an hour with your lead tech or hygienist or paralegal, read access to your systems, and one working session with you. Owners who can't free up those hours aren't ready for the roadmap, and we say so on the first call rather than deliver a thin document later.
About half of our roadmap clients come in asking for one thing and leave building something else first. The most common flip: owners ask for a customer-facing chatbot, and the audit shows the bigger leak is the phone — the average home-service business loses $50,000–$60,000 a year to missed calls. The roadmap exists to catch that before money is spent.
This is the actual sequence we run with every new client. Nothing here is proprietary magic — it's disciplined diagnosis, done by people who have seen inside a lot of service businesses.
We trace how work actually moves through your business — call to booking to job to invoice to payment. We interview the people who do the work, not just the owner, and we time the steps. The org chart says one thing; the audit finds what really happens, including the workarounds nobody wrote down.
Every system you run — ServiceTitan, Jobber, Dentrix, Clio, GoHighLevel, QuickBooks, the spreadsheet that secretly runs dispatch — gets cataloged: what data lives where, what has an API, what's clean, what's a swamp. AI is only as good as the data it can reach. This phase tells us what's reachable.
Every automation candidate gets scored on dollar impact, build effort, data readiness, and risk. Recovered revenue beats saved minutes; reachable data beats wishful data. The output is a ranked list with the math shown — not a brainstorm wall of sticky notes.
The top opportunities get sequenced into a quarter-by-quarter plan: what we build first and why, what each build needs from your team, what it should return, and how we'll measure whether it did. The first build is chosen to pay for the program — proof before ambition.
The deliverable is a written roadmap document — typically 15–25 pages, written in plain English for an owner, not a CTO. It contains:
Two things matter about this document. First, it's specific — it names your tools, your numbers, your staff's actual workflows, so a competitor's roadmap would be useless to you and yours to them. Second, it's yours. There's no lock-in clause. You can hand it to another vendor, to an in-house hire, or to nobody.
What it deliberately is not: a market-trends report, a list of AI tools to try, or a 60-page strategy document padded with definitions of machine learning. If a deliverable doesn't change a decision about your business, it doesn't go in. Most owners read the whole thing in under an hour, and the opportunity table — one page, ranked, with dollar figures — is the page that gets printed and argued over.
A composite, drawn from the pattern we see across trades clients. The owner came in asking for a website chatbot — he'd seen a competitor's and didn't want to look behind. Here's what the four phases actually surfaced.
The audit found three leaks. The office answered 71% of inbound calls during business hours and 0% after 5 PM. Estimates over $3,000 got exactly one follow-up touch, ever, from whichever CSR happened to remember. And invoices went out an average of 9 days after job completion because the techs' paper tickets sat in truck cabs until Friday.
The inventory showed ServiceTitan used at maybe 30% of capacity, QuickBooks synced but never reconciled against it, and call recordings already being captured — free training data nobody was using. The website chatbot the owner wanted would have connected to none of it.
The scoring put after-hours call coverage first (roughly $6,000/mo in missed emergency and next-day work, high data readiness, proven build), estimate follow-up second (about $11,000/mo at a conservative one extra close, medium effort), and invoice timing third. The chatbot scored ninth out of eleven — real value, wrong year.
The sequence we wrote: voice agent in month one, estimate follow-up in months two and three, invoice automation in quarter two, dashboard in quarter three once the data underneath it was trustworthy. The owner kept the chatbot on the list for year two — his call, with the math now in front of him.
That re-ranking is the product. Any vendor can build what you ask for. The roadmap exists to make sure what you ask for is worth building first.
The roadmap is available standalone. Strategy only, document is yours, no retainer required. Most standalone clients come back for the builds — but the ones who don't still leave with a plan their next vendor can execute.
Path one: we build it. The roadmap becomes the work plan for your retainer. Builds are bundled into the monthly subscription — voice agents, workflow automations, dashboards, reporting systems — never sold à la carte. The roadmap's ranked list becomes the build queue, and we re-score it quarterly as your numbers change. For a sense of what typically tops the queue, see the 12 automations that actually pay for themselves.
Path two: you take it elsewhere. Some clients use the roadmap to brief an in-house developer, evaluate SaaS tools with clear requirements instead of vendor demos, or get apples-to-apples quotes from other shops. Our guide on choosing an AI voice agent vendor pairs well with a roadmap for exactly this — once you know what you need, vendor evaluation stops being guesswork. The comparison between custom builds and SaaS tools also gets much easier to settle when you have your own numbers in hand.
We obviously prefer path one — it's our business model. But the roadmap is priced to be worth buying even if you take path two, because a strategy document that's only useful if you hire its author isn't strategy. It's a sales brochure with page numbers.
If you're under $500K in revenue. The roadmap's value comes from prioritizing among many leaks. At that size you usually have one obvious leak and not enough volume for the math to clear a retainer. Spend the money on a good SaaS tool and revisit us at $1M. Our missed-call calculator will tell you in a minute whether your phone leak alone justifies the conversation.
If your processes exist only in people's heads and you won't make time to extract them. The audit requires hours with your dispatcher, your office manager, your lead tech. If the business can't spare those hours, the roadmap will be built on the owner's secondhand version of how things work — which is exactly the document we refuse to write. Come back when you can give us the time.
If you're shopping for a magic one-click tool. A roadmap is a plan, and plans require someone to execute them. If what you actually want is "install AI, revenue goes up," no honest vendor can sell you that — and the dishonest ones are well-marketed and easy to find.
Who it is for: owners of $1M–$20M service businesses — the HVAC and plumbing companies, dental practices, roofing contractors, and law firms we work with daily — who suspect they're leaving money on the table but want the leaks measured before anything gets built. If you've already read what missed calls actually cost and felt your stomach drop, you're the reader this engagement was designed for.
Book a 30-minute call. We'll ask about your call volume, your tools, and your biggest operational headache — and tell you straight whether a roadmap would pay for itself or whether you should start somewhere cheaper. No deck, no pressure.