1. Start with your data

The most common mistake is to start by choosing a tool. A company buys Copilot or ChatGPT licences, people try them for a few weeks and then go back to their old ways of working. The tool has nothing to work with and no specific problem to solve.

AI is only as good as the data it can access. Write down where your data lives: CRM, ERP, accounting, Excel, shared drives, email. Find out what state it is in, who maintains it and where the same data is kept in several places.

Then ask people in each department where they lose time with data: what they retype from one system to another, what they search for and what they piece together into reports by hand. Typical answers: preparing quotes, sorting emails, Excel reports, customer information, retyping invoices.

2. Score tasks by benefit and complexity

For each task, write down how much time it takes per week, how many people do it and how many systems are involved. Tasks with high savings and low complexity are your quick wins. Leave complex tasks, such as forecasting or custom applications, for later.

A simple rule helps: the more systems and the more sensitive the data, the higher the complexity. For a first step, tasks in one or two systems without personal data are ideal.

3. Not everything needs AI

A large share of the savings comes not from artificial intelligence but from plain automation: transferring data between systems, notifications and generating documents from templates. Add AI where the work involves text or documents, or where something needs to be summarised, extracted or drafted. A person should always check AI outputs.

4. Run a pilot, not a big project

Choose around ten workflows and give yourself eight weeks. The first workflows should be up and running within a few days, so the team can see results. Set a success criterion for each one in advance, most often time saved. After the pilot, you know what works and you have the numbers to decide on further expansion.

5. Involve people from the start

The people who do a task every day know best where the problem lies. If you involve them in design and testing, the solution will reflect the real work and they will use it. Short training and clear rules on what AI may do and what a person checks will remove most concerns.

How much it costs

Our company review (AI assessment) costs 990 to 4,490 EUR depending on company size. A simple workflow costs 990 EUR and a standard one 2,490 EUR. You will find the details in our pricing. If you don't know where to start, an AI assessment will help.

Frequently asked questions

How long before we see results?

The first simple workflows are usually up and running within a few days. A pilot with ten workflows takes eight weeks.

Do we need our own AI specialist?

No. To begin with, you just need someone who knows the processes and a partner who will build the solution and train the team.

Not sure where to start? That's fine.

Get in touch. In a short call, we'll find out what takes up most of your time today and suggest a first step. The consultation is free and with no obligation.