There are two ways to plug AI into your tools: an open-source model installed on your premises or on servers in France or in your country, called a private LLM, or a market model such as Claude or GPT, called through an API at its provider.
Key points
- Data that must never leave the company: private LLM.
- No GPU server and a need to start fast: AI via API, on request.
- Both can coexist: the model is chosen task by task, depending on the data involved.
The two architectures
The comparison
| Criterion | Private LLM | AI via API |
|---|---|---|
| Where the data goes | Stays on your premises or on our servers, in France or in your country | Sent to the provider, under written rules |
| Hardware | GPU server on your premises, or our servers | No hardware to buy |
| Models | Open source: Mistral, LLaMA, Qwen, Gemma | Market models: Claude, GPT |
| Getting started | Installation, then adaptation to your business | Direct integration into your tools |
| AI billing | Hardware or hosting | Provider subscription, billed by usage |
Which option suits you?
Answer four questions: the tool tells you the most suitable option, and why.
Try it
Private LLM or API: the right choice for you
Your data is…
Do you have a GPU server?
You want to start…
Your tasks are mostly…
Our recommendation
Answer the four questions to see the recommendation.
Combining them
The same company can use a private LLM for its sensitive documents and AI via API for tasks with no confidential data, such as rewording a text meant for the public. What goes where is written down during scoping.
Frequently asked questions
Further reading
- Private LLM for business: the guideWhat a private LLM is, its architecture, the GPU memory it needs (calculator), what stays on your side and its current limits.
- RAG: an AI that answers from your documentsHow an AI finds the right passages in your documents, answers and cites its sources. Diagram, step-by-step demo and limits, in plain English.
- AI voice agent: how it worksHow a voice agent answers the phone day and night, takes an appointment, a booking or an order, and hands over to your team when needed.
In 30 minutes, we look together at whether a pilot makes sense for you.