EN English DE Deutsch
Request an Initiative Check Initiative Check
Menu
EN English DE Deutsch
Request an Initiative Check

Initiative · Project

Introducing AI

An AI pilot is quick to build. Running it isn't.

Finding useful AI applications, preparing data and rules, building solutions and enabling employees to use them.

The situation

“We want to use AI, but properly.”

Some of your team have long been using AI, with private accounts and real company data. You're wondering what's actually allowed. And the chatbot from the last trial is still running, but nobody asks it anything.

Ideas are plentiful: writing quotes faster, pre-sorting enquiries, summarising meeting notes. The open question is which ones pay off, which data is good enough and which rules apply.

Where it often goes wrong.

  1. A tool looking for a problem

    First an AI solution is bought, then a use is found for it. Yet AI is a tool, and you choose tools to fit the task.

  2. Data nobody has maintained

    AI is only as good as the data behind it. Outdated price lists and duplicate master data produce very convincing wrong answers.

  3. Rules only after the incident

    Without clear rules, confidential data ends up in third-party services. Write the rules only after the first incident, and you'll write them under pressure.

  4. Access handed out, job done

    Handing out licences takes no time at all. Without practice on real tasks, soon only the few who were keen anyway still use them.

  5. Costing only the test run

    In a test run, AI costs little. In operation, fees per request, maintenance and checks come on top. Then some use cases no longer pay off.

Which disciplines work together here.

3 of 13 disciplines, one team, one point of contact. Every service has a clear role, and you decide which ones we take on.

You decide the scope.

A first step can be assessing your ideas: what pays off and what doesn't? Equally possible is a single use case taken all the way into operation.

Or you start with rules for the AI your team is already using. That's often the most urgent part.

Good questions. Straight answers.

Are we even allowed to use AI with customer data?

Often yes, under certain conditions. What matters is the purpose, the contract with the provider and where the data is processed.

The legal assessment is done by a law firm that we appoint on your behalf. We implement what it sets out, including the duties under the AI Act.

Will AI replace jobs in our company?

Rarely entirely. Usually AI takes over individual steps, such as pre-sorting or summarising, and people check and decide.

What happens with the time freed up is your decision. We plan the transition so that nobody is presented with a fait accompli.

Do you use a particular AI model?

No. Which model fits depends on the task, data protection and cost. We choose accordingly and build it so you can switch later without starting from scratch.

Should we start with a chatbot?

Not necessarily. Chatbots are visible, but they don't always bring the biggest benefit. Often the invisible work pays off first: reading documents, sorting enquiries, matching data.

Sometimes even simple automation without AI will do.

Do we need our own data specialists for this?

Not to start with. Building and running can sit with us.

In the long run, though, you need people in-house who check results and spot new use cases. We train your team for that, even if operations stay with us.

Proof first. Contract second.

In the free Initiative Check, we listen for 45 minutes and ask the questions that usually only come up after the contract is signed. Then you get a short written assessment: what we see, how we would go about it and roughly what it would take.

No obligation. And if we're not the right fit, we'll tell you.

Request an Initiative Check

Or write to us at hello@intoops.com.