AI consulting for home service businesses that ends in working systems

If you hire an AI consultant for a home-service business, the engagement should end with something running inside your field software within about 30 days: an agent that logs missed calls, chases quiet estimates or builds the owner's daily report. If it ends with a slide deck and a list of tools to try, you paid for advice, and the work of making it real is still yours.

Both kinds of consultant exist. They charge for very different things. This page explains what each one produces, what a good first month looks like, and the questions that separate them.

Two kinds of AI consultant

Advice-only consultants study your business and tell you what to do. You get interviews, a readiness assessment, a roadmap and maybe a vendor shortlist. This is useful if you have staff who can carry it out. Most trades companies don't. The office manager is already running payroll, permits and the phones.

Implementation consultants do the advice part and then build it. They connect to your FSM (ServiceTitan, Housecall Pro, Jobber), your CRM, your phone system and your books, then install agents with specific jobs and a person approving what matters. You can see the difference in how an AI implementation project runs.

The labels blur on purpose. Plenty of firms say "AI implementation" and deliver a deck. Plenty of software vendors say "consulting" and mean setting up their own product. Gartner estimates only about 130 of the thousands of agentic AI vendors are real, and calls the rest "agent washing": chatbots and old automation with a new name.

Why trades owners are hiring help now

Small companies are falling behind on actual use. The Census Bureau's business survey found AI use rose among firms with 20 or more employees but didn't change much among firms with fewer than 20, between December 2025 and May 2026.

That doesn't mean small owners aren't trying. The U.S. Chamber of Commerce found that 58% of small businesses say they use generative AI, up from 40% a year earlier. Most of that is someone pasting emails into ChatGPT. The gap is between using a chat tool and having AI do a job inside the business, which is exactly the gap a consultant gets hired to close.

The other pressure is labor. BLS expects about 40,600 openings a year for HVAC technicians over the next decade. Owners can't hire their way out of office work when they can't hire enough techs either.

What a good 30-day engagement produces

Here is what we think the first month should deliver for a 10 to 80 person service company. Hold any consultant to something like it.

Week 1: the map. Interviews with the owner, a CSR, a dispatcher and whoever does billing. Read-only access to the systems. Out of it comes a written list of where work drops: calls that never became jobs, estimates with no follow-up, jobs finished but not invoiced, reviews with no reply. Each item is ranked by how often it happens and how hard it is to fix.

Week 2: the design. For the top one or two items, a one-page spec per agent. What it watches. What it can write. What it drafts for a person. What it never touches. Who gets the alert when it fails.

Weeks 3 and 4: the build. The first agent goes live in your real systems on a small slice of work, with a person approving every action. Drafts get compared against what your best CSR would have sent. Permissions widen only after the drafts are trusted. We explain that approval model in human-in-the-loop AI.

End of month: the handoff. Documentation of what was built, where the logs are, how to turn it off, and who maintains it from here.

If a consultant can't describe their first month at this level of detail, they probably haven't done it before.

What it costs and how to compare

We won't quote other firms' prices. They vary too much and few publish them. Two benchmarks help you judge a proposal.

First, the in-house alternative. The BLS median for software developers is $135,980 a year, before benefits and recruiting. For context, the median for management analysts, the category that includes consultants, is $101,860. An advice-only engagement buys management-analyst work. An implementation engagement should look more like engineering work, and the output should match.

Second, the failure rate. Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027 because of rising costs, unclear value or weak risk controls. A cheap project that gets canceled costs more than a scoped one that keeps running. Ask how the consultant measures whether the agent is working, and who pays to fix it when it isn't.

Questions to ask before you hire

• What will be running in our software at the end of the first month?
• Which of our systems have you connected to before, and what did you do there?
• What will you refuse to automate? (A good answer names pricing on messy jobs, upset customers and anything that makes a promise.)
• Who gets the alert when an agent fails at 2 AM?
• Can we see a live system, not a demo?
• How do you handle risk? NIST's AI Risk Management Framework gives a vocabulary for this (govern, map, measure, manage). Nobody needs a 50-page policy, but the consultant should be able to explain how they map and measure risk in plain words.
• What do we own if we stop working together? Accounts, data and documentation should stay with you.

Red flags

• A long list of tools before anyone has looked at your call logs or job board.
• Pricing per "AI agent" with no description of what each one does.
• Promises of a specific revenue lift before they've seen your numbers.
• No plan for after launch.
• They want to replace your FSM. Good implementation works inside the software you already pay for. See why we build inside existing software.

Where Cortex fits

Cortex is an implementation partner. We do the consulting part (the audit and the design) and then build and run the agents inside the software you already use. The path is the same every time: audit, design, build, deploy, manage. You can read each step on how we work. If you are weighing a part-time team against your own hire, our guide to a fractional AI team runs that math.

We don't publish fixed prices because the work depends on your systems and how many workflows you want handled. We scope after we've seen the work.

The first step is free: the AI Operations Audit, a 30-minute walk through your software and workflows. You leave with a ranked map of where AI fits and what we would leave alone, whether or not you hire us.