What does AI consulting cost a small firm in Switzerland?

At PMBRTN, an AI Opportunity Audit costs CHF 750 to 1,500, a first automated workflow starts from CHF 1,500, and a custom AI tool starts from CHF 5,000, each at a fixed price. What moves a project up or down is the number of systems involved, the state of your data and how much checking the output needs, not the AI itself.

I publish my prices because a managing partner should be able to budget before the first call. When I looked at what comes up for this question, the guides were almost all written for American or British firms in dollars or pounds. This is the Swiss version, in francs, for a firm of 5 to 100 people.

Three levels of engagement

1. An audit: CHF 750 to 1,500

One to two weeks. You get a written map of where time goes, the two or three workflows worth automating first, what not to automate, and a fixed quote for the first build. This is the cheapest way to avoid the most expensive mistake, which is automating the wrong thing. How the audit works.

2. A first workflow: from CHF 1,500

Two to six weeks. One workflow, live and documented: for example a shared inbox where AI drafts replies for review, or invoices and forms whose data lands in a spreadsheet without retyping. What can be automated.

3. A custom tool: from CHF 5,000

Four to eight weeks or more. A small application for a job no product does well for your firm: a quoting tool on your own pricing rules, a knowledge assistant over your templates and policies, or a dashboard joining several systems. Examples.

An optional monthly retainer covers iteration and new workflows afterwards. None of it is required.

What moves the price

• Number of systems. One inbox and a spreadsheet is a small job. Email, a practice system and a document store working together is a bigger one.
• State of the data. Consistent templates and folders are quick. Fifteen years of ad hoc file names are not.
• How much review the output needs. Anything that reaches a client needs a checking step designed in and tested. In a law firm or Treuhand, that is most things.
• Confidentiality requirements. Choosing a service with the right contract terms and setting access properly takes time, and it is time well spent.

Running costs to budget separately

Consulting is a one-off. Some workflows also need software you pay for each month, billed by the vendor rather than by me:

• A business AI plan or licence, such as Microsoft 365 Copilot (Swiss prices and rollout order) or a ChatGPT business plan. For client data, use a business plan rather than personal accounts: OpenAI does not train on business plan data by default, and Microsoft states the same for Copilot. [1, 2]
• Usage-based AI costs for automations that call an AI model directly. I estimate them in the quote.
• Hosting, only if a custom tool needs it. Listed in the quote.

Where your existing tools can do the job, that is what I use, so there may be no extra running cost at all.

How to judge whether it is worth it

Work it out on your own numbers, not on a vendor's percentage claims. Take the hours a task costs each week, multiply by the loaded hourly cost of the people doing it, and compare a year of that with the one-off price plus a year of running costs.

For illustration only: a task that takes three hours a week of someone costing CHF 80 an hour is about CHF 12,000 a year. If a CHF 1,500 workflow removes half of it, it pays for itself within a few months. If the task only takes an hour a month, it probably is not worth automating, and a good audit will tell you so.

Red flags when buying AI consulting

• Open-ended "strategy" retainers with no defined deliverable.
• Replacing your systems when the work could be done on the tools you already have.
• No plan for confidentiality beyond "the tool is secure".
• No documentation, so only the consultant can run what was built.

Every PMBRTN project is fixed-scope and fixed-price, built on your existing tools where possible and documented in plain English. See the pricing page.