What a half-day AI assessment actually looks like, hour by hour

People ask us what actually happens during a half-day assessment. It is a fair question. “AI assessment” can mean a slide deck, a sales pitch, or a survey you fill out while someone waits. Ours is none of those. It is four hours on site, mostly spent watching how work gets done, and it ends with a short written memo and one small fix running before we leave.

Here is the plan, hour by hour. The times are approximate. Every shop is different, and we adjust on the day. But the order does not change much, because each hour sets up the next one.

Before we arrive

A few days ahead, we send our one-page worksheet, Where the hours go. It asks you to list the recurring tasks that eat the week, who does each one, how often it happens, and how long it takes, then add up the hours. It takes about 15 minutes with your team.

If you fill it out, the first hour goes faster. If you do not, that is fine. We have watched enough teams work to find the same answers on site. The worksheet mostly helps us avoid spending your time on questions you could have answered in writing.

We also sign an NDA before looking at anything. We will be reading your spreadsheets and watching your screens. You should not have to wonder where that goes.

Hour one: the owner conversation

We start with you, or whoever runs the day to day, for about forty-five minutes. Not a presentation. A conversation, with a notebook.

We are listening for three things. Where does money leak? Where do people work late? Where does the same question get asked every week? Owners usually know the answers in their gut and have never said them out loud in one sitting. Saying them out loud is useful on its own.

We also ask what you have already tried. Most businesses we talk to have bought at least one tool that nobody uses anymore. That history tells us a lot about what will and will not stick with your team.

By the end of the hour, we have a list of five to ten candidate tasks. We have not judged any of them yet. That comes later, after we have watched them.

Hour two: watching the work

This is the part that makes the assessment worth doing, and it is the part that a phone call or a questionnaire cannot replace.

We sit next to the two or three people who do the tasks on the list. Fifteen minutes each, at their desk, doing the job the way they normally do it. We do not ask them to explain it. We ask them to do it while we watch, and we ask questions only when something surprises us.

What we are watching for is specific. Where does information get typed twice? Where does someone open four windows to answer one question? Where does a person wait on another person for something a system could have sent? Where does a process exist only in one employee’s head?

People are often a little nervous about this part. It passes quickly. Most people like showing someone the workaround they invented, and most of the best findings come from exactly those workarounds. The employee who built a side spreadsheet to track something the main system cannot has already told you where the gap is.

We also pay attention to what we do not see. If the owner described a task one way in hour one and the person doing it does it a different way, that is a finding. Usually it means the process needs to be written down before any tool gets pointed at it. We wrote more about that in our post on fixing the process before buying software.

Hour three: ranking what we found

Around the midpoint, we step away for a while, usually to a spare desk or the truck, and turn the notes into a ranked list.

Each candidate gets scored on four things: how many hours a week it eats, how many people it touches, how much of it is repetitive enough to hand off, and how hard the fix is. The first three push a task up. The fourth pushes it down.

The ranking matters more than the list. Anyone can produce a list of twenty things that could be improved. The useful question is which two or three are worth your money first, and which ones should wait or never happen. We put “do not bother” items on the memo on purpose. Knowing what to skip saves as much as knowing what to do.

We also sort the fixes into three buckets. Some are process fixes that need no software at all. Some are connections between tools you already own. Some are genuine automation or custom build candidates. The order of those buckets is the order of cost, lowest first. We explain what each bucket tends to cost as a market range, labelled as such, as of the month we visit.

Then we pick the one fix we will build before we leave. The criteria are simple. It has to be small enough to finish in about an hour, safe enough that it cannot break anything important, and visible enough that the person it helps notices on Monday.

Hour four: the quick win and the walkthrough

The last stretch splits in two.

Most of it is building the fix. Typical examples: a template that drafts the follow-up email your estimator retypes six times a day. A rule that moves incoming job photos into the right folder automatically. A short form that replaces the sticky note someone hands to the office manager. None of these is impressive on its own. That is the point. It is a proof that something in your shop can change in an hour, and that the change can be done without a new platform.

We build it with the person who will use it, not for them. If they cannot change it themselves after we leave, it is not finished.

The rest is the walkthrough with you. We go through the ranked list out loud, top to bottom, and answer questions. We are honest about which items we could help with and which ones you could do yourselves, or with a tool you already have. If the honest answer is that nothing on the list is worth a bigger project, we say that, and the half day is still worth the memo and the one working fix.

What you have at the end

Three things, all in your hands the same day.

A ranked memo, usually two or three pages. Every item has a short description of the problem, the hours at stake, the right fix, and whether to do it next or skip it. You can read a sample findings memo to see the format. The sample is illustrative, not from a client, but the structure is exactly what you would get.

One change already working, with the person who uses it knowing how to adjust it.

A plain recommendation about what to do next. Sometimes that is a two-week sprint on the top item. Sometimes it is a half day of team training because the tools are fine and the habits are not. Sometimes it is nothing, and we tell you so.

There is no pitch at the end and no follow-up sequence. If you want to take the next step, the memo tells you what it is and roughly what it costs. If you want to hand it to someone else, you can.

Why a half day and not a week

Because a week is mostly wasted on a business with fewer than a hundred people. The patterns are visible in hours, not days. Stretching the visit out mostly adds cost and meetings.

A half day forces us to focus on what is actually costing you time, and it forces us to prove something on your floor instead of in a document. That constraint is the product. The full scope and price are on the opportunity assessment page.

Common questions

Do we need to prepare anything before the half day?

Very little. We send a one-page worksheet ahead of time that asks you to list the recurring tasks that eat the week, who does them, and how long they take. Fill it out if you can. If you cannot, we figure it out in the first hour.

Who needs to be there?

The owner or manager for the first and last hour, and whoever actually does the work in between. We do not need the whole team in a room. We need access to two or three people at their desks for fifteen minutes each.

What if you do not find anything worth doing?

Then the memo says so, you still get the one quick win, and we tell you to stop there. We would rather say that plainly than invent a project.

Does the assessment require us to buy anything?

No. Most of what we recommend first uses tools you already pay for, such as QuickBooks, Google Workspace, or Microsoft 365. If something does need a new subscription, the memo says what it is and what it costs as a market range.

What does it cost?

From $750 for the half day, as of August 2026. That includes the visit, the ranked memo, and the one fix we build on site. There is no obligation to do anything further with us.