Giving AI a place in your organization

Many organizations are experimenting with AI. A few colleagues use a chat assistant, someone automates a process. Moving to a standard way of working that the whole organization uses turns out to be harder. That is where we can help.

We start with the questions that really matter. Which problem do you want to solve, and for whom? Which systems and data does AI need for that, and which should it stay away from? How do you handle access rights and privacy, and where is your data processed? And how do you bring your colleagues along, so that it actually gets used afterwards?

It helps that we work with AI every day ourselves. We see for ourselves where it works well and where you need to set limits. A recent example is our support process. An employee has an AI agent collect the open customer mails. For each question, the agent creates a support issue, works out what is going on in the application, the logs, known issues and the code, and prepares a draft reply.

The agent works within clear limits. It has read-only access to the application and the logs. The reply stays a draft that an employee reviews and sends themselves. If the code needs to change, a development issue is only created after approval. Limits like these matter as much as the technology, and they are exactly what we like to think through with other organizations.

Diagram of the support flow. The AI agent follows written procedures, reads the mailbox, the application, the logs, the issues and the code, and writes a draft and a support issue. An employee sends the reply and approves new development issues. — Our support process as an example, from customer mail to draft reply. Click the diagram for a larger version.

Want to know what AI could mean for your organization, and how to approach it carefully? Get in touch.