Your Own ChatGPT, on Your Own Servers

01 What is a private LLM?

A large language model that runs on machines you control, in your office or your own cloud account, instead of a provider's service. What your team types, and the documents it uses, go to your own server, not to OpenAI, Anthropic or anyone else. We run open models with Ollama for this.

02 Is a private LLM as good as ChatGPT or Claude?

It depends on the task, and the pilot is how you find out. We test open models on your own questions, and compare them with a hosted model where your rules allow, so you see the difference in answer quality before you decide where the model runs.

03 Can we have a private ChatGPT for our team?

Yes. We set up a chat app your staff sign in to, answering with an open model on your servers. It can search the documents you choose, and each person reaches only what their role allows. Nothing they type goes to an outside AI provider.

04 Where does our data go?

Where you decide. With hosted models from OpenAI or Anthropic, data is processed by that provider. With open models on your own servers or cloud, documents and questions stay inside your network. Many teams use both, with the private model for the data that must stay in.

05 What hardware does a private LLM need?

It depends on the size of the model and how many people use it at once. The pilot runs your chosen model on your own questions and measures what it needs. You then size the hardware on what was measured.

06 On-premise or in our own cloud?

Either. On-premise means servers in your office or data centre. Your own cloud means machines in a cloud account in your company's name, inside your private network. Both keep the data off any outside AI provider. The choice is usually about your rules and your hardware.

07 Which open models do you deploy?

Models that Ollama runs, such as Llama, Qwen, Gemma and Mistral, chosen per task on answer quality, speed and the hardware they need. Ollama can also load a model that was trained further, such as one you have fine-tuned. We test more than one on your questions before choosing.

08 Private LLM or fine-tuning: which do we need?

A private LLM is about where the model runs. Fine-tuning is about training it on your examples. Most teams start with a private open model, clear instructions and search over their documents. Fine-tuning comes later, if the model still misses your format or style; our LLM fine-tuning page covers when.

09 Can our existing apps use the private model?

Yes. Ollama, the tool that runs the model, accepts part of the same connection format (API) that OpenAI uses. So software already written for OpenAI can often use your private model with small changes. We put sign-in and a usage log in front of it, so only your apps and people reach it.

10 How much does a private LLM deployment cost?

Three things drive the cost: the model size and hardware, how many people and apps use it, and how much document search and sign-in it needs. Tell us what you want it to do, and within two working days we come back with an approach, a rough timeline and a rough number.

11 How long does it take?

It depends on the hardware, the data and how many systems it connects to. We start with a pilot on one machine, so your team sees answers before a full rollout. We agree the timeline at the discovery call.

12 Who owns the deployment?

You do. The model runs on your machines, the code goes into your own code repository, and the cloud account is in your company's name. You get the documentation and a handover, so your team can run it yourselves.

13 What happens after launch?

We monitor the deployment, keep it updated and test new open models as they arrive, swapping one in only when it does better on your test set. You can also run it yourselves.

14 Do you work with UK and US clients from India?

Yes. We sign an NDA before we start, on request, and work only in the accounts you invite us into. The code goes into your own code repository, in an account in your company's name.