Skip to content

What is local AI?

Local AI runs on your own hardware instead of at a supplier in the cloud, so your data stays in-house.

In short

With local AI, the model sits on a server or workstation that you manage yourself. Questions and documents do not go to an external service.

When to choose it

  • You work with confidential data, such as case files or financial records
  • You want fixed costs instead of paying per question
  • You do not want to depend on the terms of a single supplier

What you need

A machine with enough memory, a language model that fits on that machine, and software that makes the model available to your applications. The larger the model, the more memory it needs. A suitable graphics processor can speed up processing. A suitable graphics processor can speed up processing.

Not just text: speech recognition, such as transcribing meetings, can also run entirely locally.

Would you like advice on local AI? Request a free advice conversation.

Updated on

AI

Personal data in the AI cloud: who is watching?

A conversation with an AI service in the cloud is not private by default. A case in Canada shows what a provider can see and share. What does that mean under the GDPR, in the United Kingdom and in Switzerland?