How long does it take to set up AI estimating in a shop?

About a week for Mechanica. Most of that week goes to learning how your shop quotes, and your team grades practice estimates before any draft reaches a real customer. Your techs keep writing inspections the way they do today, and the only access the tool needs is a login with a new advisor's permissions, or fewer if you set it that way.

Key takeaways

• Setup is mostly the AI learning your pricing, which means your labor rate, markups, canned jobs and preferred brands.
• Techs keep their current inspection habits, and a photo of a paper sheet works.
• Practice estimates get graded by your advisors before anything goes live.
• Messy canned jobs and an unsettled parts matrix are what slow a setup down.

What happens during the setup week?

Four stages, and the shop does very little of the typing in any of them.

Access. The tool gets its own login in your shop system with the permissions you would give a new advisor. It drafts on the repair order, and that is the whole footprint.

Your pricing. Mechanica loads your labor rate and your markups, then your canned jobs and the brands you prefer to install. This is the part that makes an estimate read like your shop wrote it. A brake job should come back priced the way you already sell brake jobs.

Practice estimates. The AI drafts estimates on your real tickets while nothing is live. Your advisors grade them line by line and mark what is wrong. Those marks show where the drafts miss how your shop prices, and Mechanica gets tuned to close the gaps before launch.

Going live. Drafts start showing up in the review queue on the repair order, and your advisor approves each one before it goes to the customer.

What access does it need?

As little as possible, which the Federal Trade Commission treats as the default for anything that handles customer information. Its guide to protecting personal information says to "Follow the 'principle of least privilege.' That means each employee should have access only to those resources needed to do their particular job."

For an estimating tool, the job is reading findings and drafting estimates. A regular user login covers that. Administrator access stays with your own people, and the tool should have its own named user instead of borrowing an advisor's password, so you can shut it off in one step if you ever need to.

We go deeper on this in is it safe to let AI software into your shop system, including the questions worth asking any vendor before you connect it.

What do you need to hand over?

Information you already have, most of it sitting in your shop system today:

• Your labor rate, and any rate that differs by job type.
• Your markups, ideally as a parts matrix so they apply the same way on every line.
• Your canned jobs, the templates your advisors already use for common work.
• Your preferred brands for the parts you install most.

If some of that lives in an advisor's head instead of in the system, setup is when it gets written down. That turns out to be useful on its own, because the next advisor you hire needs the same information.

Do techs or advisors need training?

Techs keep doing what they do now. The AI reads the findings they already write, and if they write on paper, a photo of the sheet works. Notes that name the component and its position price cleaner, and what a clean inspection note looks like shows the format, though nobody has to adopt it before day one.

Advisors learn one new habit. Instead of building an estimate from a blank screen, they open a draft, read it and approve it. The judgment stays with them, and so does the conversation with the customer.

What are the practice estimates for?

They are where the setup earns its week. Every correction an advisor makes during practice points at something about how your shop prices, like a labor time you always adjust or a job you always sell with a second line.

Grade them the way you would grade a new advisor's first estimates. Check that each job matches the finding and the labor fits the vehicle. Then check every part against what you would order and what you would pay. A practice draft that misses gets corrected before the tool goes live.

Nothing from a practice estimate reaches a customer. The point is to find the gaps while the stakes are zero.

What can you do before the week starts?

A little preparation shortens the week, and every item on this list pays off even if you never install anything.

• Write down your labor rates. One number for most shops, a few for shops that charge differently for diagnosis or for certain work.
• Pull your canned job list and mark the ones your advisors use in a normal week. The rest can wait.
• Write your markup rules in one place. If two advisors would describe them differently, settle it before setup instead of during it.
• Pick the advisor who grades the practice drafts. Your most experienced writer is the right choice, because the drafts get tuned to match how that person prices.
• Set aside a handful of last month's tickets with a mix of work on them. Real tickets make better practice than made-up ones.

Who at the shop should be involved?

Two people, for short stretches. The owner decides the pricing rules the drafts follow, because rates and markups are an owner's call. The lead advisor grades the practice drafts, because that person knows what a correct estimate from your shop looks like.

Techs stay out of it unless a finding needs explaining. Their part of the job does not change. If a practice draft keeps misreading one kind of note, that is the moment to ask the tech who wrote it what they meant, and the answer usually improves the note as much as the draft. A short conversation at the bay settles more than a long one at the counter.

What slows a setup down?

Usually the shop's own records.

Canned jobs that are out of date. A template built three years ago, with parts that have since been superseded and labor that nobody bills anymore, will get copied faithfully until somebody fixes it. Our guide to auditing your canned jobs covers what to check.

Markup decided at the counter. When each advisor marks up parts their own way, there is no single rule to load. Settling it first with a parts matrix template makes the AI consistent, and it makes your advisors consistent too.

Neither of these is extra work created by the software. Both were already costing the shop money on every ticket priced by hand.

How do you know it is ready to go live?

When your advisors would send the practice drafts with light edits. A draft is ready when it holds the pieces the FTC tells car owners to expect in a written estimate, which "should identify the condition to be repaired," along with the parts needed and the anticipated labor charge, all priced the way your shop prices.

The approval step stays in place after launch. California's repair statute says "No work shall be done and no charges shall accrue before authorization to proceed is obtained from the customer," and that authorization is still your advisor's call on every ticket. The AI drafts, and your advisor decides what goes out.

What changes on the first live day?

The tech finishes the inspection the same way as yesterday. A drafted estimate shows up on the repair order in the review queue. Your advisor reads it, adjusts anything that needs it and sends it to the customer.

What your advisor stops doing is building the estimate line by line from a blank screen. The time that used to go into lookups goes back into the phone calls that sell the work. If you want to see the pricing logic before you commit to anything, try the estimate builder, or read can AI write an estimate from the tech's inspection notes for what the AI needs to get it right.

More on putting AI to work at the counter is on the blog.