Service
Local AI
Local AI is an internal Kod13 project. We use it as a foundation for a separate assistant tailored to a customer’s team and running in an environment they control.
Who it is for
Teams that want to search internal documents or use AI without sending their data to a public chat service.
What it covers
- Scoping the tasks, data, and limits
- Choosing a model for the available hardware
- A private web chat in the customer’s environment
- Configuration and a short team handover
What you get
- The environment can remain under the customer’s control
- The model is selected for the actual hardware
- The implementation fits the team’s specific work
How we do it
Kod13’s internal installation is a development base, not a public product with user sign-ups. A customer gets a separate implementation shaped around the task, available hardware, and data access.
The model is not an archive and does not train itself on files. Before work starts, we define what the assistant should do and which limits matter to the team.
How an engagement runs
01. Tasks
We define what the assistant should and should not do.
02. Hardware
We check the machine and select a suitable model.
03. Setup
We run a private chat in the customer’s environment.
04. Handover
We show the team how to use it and document its limits.
Questions on this topic
Do you expose your internal AI address as a public service?
No. Our internal installation is for our own use. Each customer gets a separate implementation.
Do we need a graphics card?
Not always. It depends on the task, model, and the hardware available.