What is local AI?
Local AI is AI that runs on the organisation's own computers or servers, so questions and documents do not need to be sent to an external service.
With a public AI service, every question travels to a server somewhere else. With local AI the model is where you are. The question, the documents and the answer stay in your environment.
Local AI can be connected to the internet or fully isolated. What sets it apart from a cloud service is where the computing happens.
It gives control over the data but needs someone to run it. iSearch can run fully offline at your site or in a dedicated environment with us.
How it works
Local AI requires the model to be in your own environment, usually an open model, and servers able to run it. Documents are read in and indexed in the same place. When someone asks a question, the search, calculation and answer all happen without anything being sent to an outside service.
Why it matters
Local AI makes it possible to use AI on information that may not leave the organisation, such as patient data, confidential records or trade secrets. It also simplifies questions about third-country transfers, since no such transfer takes place.
Consider a municipality that wants its social services staff to query their guidelines and templates. With local AI the material can be used without any data leaving the municipality's network.
What to keep in mind
Local AI moves responsibility for operations, updates and capacity to you or your partner. Plan for how many people will use it at once.
A local model may be smaller than the largest services. For questions about your own documents that often matters less, but test with your most common tasks.
What it means for you
Local AI is often the shortest answer to questions about confidentiality and data protection. Weigh it against who will run it.
Common questions
Is local AI enough for our needs?
For questions about your own documents, the material often matters more than the size of the model. Try it with your own questions and compare.
What does it take to run AI locally?
Servers with enough computing power, often with GPUs, and someone responsible for operations and updates.
Is local AI the same as on-premises?
The two are closely related. On-premises refers to the whole software installation. Local AI stresses that the AI model itself runs on your side.
Related terms
Read more
All terms Updated