Skip to main content
Webryk Studio
FR
Back to services

AI service · Diagnosis

SMB AI audit
Know where AI is actually worth it.

I go through the tasks that keep coming back, the content you already have, and where the risk actually sits. You leave with one or two simple uses worth testing first. Buying more software is not the point.

From $249

One-time. The $99 web presence check looks at what the public sees; this one looks at your internal tasks, your data and your risks.

Problems solved

You have the tools, but nobody agreed on how they get used
No one has said which sensitive data is allowed near an AI tool
You cannot tell which use is worth trying first

Possible deliverables

Prioritized use list
Risks and limits
Impact vs effort
Simple action plan

How it runs

Every mandate follows these steps, in this order, and none of it depends on one particular tool.

  1. 01

    A normal week, described

    I want to know which tasks come back every week, who does them and how long they take. The list covers real actions, not the software you own.

  2. 02

    Your content and your data

    What already exists as documents, standard answers and procedures. And above all, what must never leave your walls.

  3. 03

    Sorted by impact and by effort

    Plenty of AI ideas cost more time than they free up. When that is the case, it is written plainly in the document.

  4. 04

    One or two things to try

    You leave with a plan you can test without me, and with what to watch to know whether it is working.

What the mandate covers

The scope is settled before anything starts, and none of the below arrives as an extra once it is agreed.

The scoping conversation

You walk me through a normal week: what comes back, who handles it, how long it takes. It is the one point where I ask a lot of questions.

The inventory of repeated tasks

Written down, with who owns each one and how often it comes round. A task nobody can describe cannot be automated, and writing it down sometimes solves it outright.

One or two scoped experiments

With what to watch so you know whether it is working. You can run them without me, and that is deliberate.

The document, which stays yours

We go through it together before you leave with it. It is written to be useful whether you carry on with Webryk or not.

When this is the wrong mandate

If the goal is to replace staff, this is the wrong mandate: the uses I keep remove tasks, not people. And if no repeated task is documented yet at your company, the audit comes too early. Start by writing down what you do, and we will talk after that. And it is not a review of your website: that is the $99 web presence check. The AI audit looks at how you work, not at how you look online.

Privacy and simplicity

I write down which data the tool can see, which answers it is allowed to give, where it stops, and at what point a human takes over. It is the dull part of the mandate, and it is what separates a tool you end up using from one that only looks impressive.

What people ask me before we start

Which AI tool is used, and who chooses it?
The tool is chosen with you: either the one you already use, or the one I recommend for your case. It is named in the contract, along with what it sees and what it does not, because Law 25 requires a written agreement with the provider when personal information leaves Quebec, and an assessment before it goes. I do not impose a tool and I do not swap it without telling you. The processing happens outside Quebec, usually in the United States, because that is where the providers are: it is written into the contract rather than left to be guessed, and you get thirty days notice before any change of provider.
What do I get at the end?
A document listing the possible uses sorted by impact and by effort, the risks tied to your data, and an action plan for the first one or two experiments. We go through it together before you leave with it.
How is this different from the $99 web presence check?
They are two separate things. The $99 check looks at your website, your Google listing and your social pages, and tells you what is missing online. The SMB AI audit looks inside: which tasks come back every week, who does them, which data must not leave the building. A client who wants to know where they stand publicly takes the check; a client who wants to know where AI would save them time takes the audit.
Do I need to already use AI tools for this to be useful?
No. The audit starts from your tasks, not your tools. A company that has tried nothing is often simpler to scope than one already running five subscriptions in parallel.
What if the conclusion is that AI is useless for us?
That is a possible conclusion and I say so when it applies. You still keep the list of your repeated tasks and the state of your data, both of which stay useful without AI.
Do you install the tools afterwards?
Not as part of the audit. Setup, subscriptions and integrations are scoped separately, and the document is written so you can move without me if you prefer. An audit whose only possible conclusion is "hire me" is not worth what you paid for it.
Do we need to be a large company for this to pay off?
No, and often it is the reverse: the fewer people to consult, the faster the picture comes together. Repeated tasks exist at every size. Where the audit is useless is when nothing is documented yet.
Do I have to show you confidential data?
No. The data map deals in categories, not contents: we note that client files exist, not what is in them. That is precisely what makes it possible to write down which categories must never leave.

Let’s see whether AI truly has a place in your case.

Schedule a call