Smart AI for automation

We arrange that before anything is deployed. We determine which data may be processed, choose tools with appropriate security and clear agreements on storage, and make sure confidential information is not used to train the supplier's AI models. If that cannot be arranged properly for certain data, that data does not go into AI.

How many applications you want, which systems AI needs to work with and how much training your team needs: those three determine the scope. We deliberately keep the AI scan and a first pilot manageable. After the introductory call we give you an indication of the effort for those first steps, along with the planning. We work on an hourly basis and invoice afterwards, so you only pay for the time we actually spend.

Yes, we even recommend it. We start with one application, a small group of employees and a clear goal. If it works, we expand. If it doesn't, you find out quickly and without a large investment.

Hardly ever. For most SMEs, existing AI models and tools are enough, set up with your own documents, ways of working and ground rules. Training your own model is costly and rarely necessary. Only if existing tools really fall short do we look further.

We choose per business, with no preference for any brand or provider. We prefer to start with what you already have, such as the AI features in your office software, CRM or accounting package. Where needed, we add specialised tools or integrations. The choice depends on your tasks, your data and your security requirements. We may also use different models for different tasks.

MCP stands for Model Context Protocol: an open standard that lets an AI chat work together with other software. An MCP server is the link between the chat and, for example, your CRM, accounting or planning. It lets employees look up and process data from the chat, with the same rights as when they log in themselves.

Only if you want it to, and only where you want it to. For each system we set out what AI may read and what it may change. We often start with read-only access. For each write action we can also decide whether human review is required.

The honest answer: as a business owner, that is largely up to you. AI takes work off people's hands, so less capacity is needed for those specific tasks. That does not make an employee redundant. There is often work elsewhere in your business that AI is less suited to and where that same employee makes the difference. Our aim is to relieve your people: AI takes over repetitive groundwork, leaving more time for other tasks. And if your business grows, your team can handle more work without the workload growing at the same rate.

By involving them from the start. Employees know best where the time goes, so they help shape the AI scan and test the pilot. Then follow practical training sessions using their own tasks as examples. What people see working well for themselves, they will also use.