Plenty of businesses hold material they would never paste into a public chatbot: client files, contracts, customer lists, draft numbers. AideaMaker Private AI is our answer for that kind of work. This guide says plainly what it is, what happens to what you send, and where its promises stop.

What Private AI is

AideaMaker Private AI is an AI service we run ourselves. It is OpenAI-compatible, so developers can point an existing chat app at it, and it gives you a choice of model from an approved list: Laguna and Qwen3.8 Flash Next. You make a free account to try it, and you can use it directly from your own apps or run your AideaMaker software on it. See the Private AI services.

What happens to what you send

Your prompts and answers pass through our API gateway without being written to our databases, logs or backups. We keep usage records only: which key made the call, which model answered, how many tokens went each way, how long it took, whether it worked, and the calling address and software name. Those records let us bill, limit and run the service, and we keep them for up to 400 days.

That is a promise about the Private AI service, not about everything around it. If you run an AideaMaker product on top of Private AI, that product keeps what it is built to keep: a hosted agent, for example, keeps its conversation history in your own portal so you can read it. We say so on the product's pages instead of implying that everything around the service keeps nothing.

Pick your model

Private AI runs two models today, and you choose between them: Laguna and Qwen3.8 Flash Next. Try both free with your own prompts before you decide, then pay per use, or rent a model for yourself. A rental is prepaid for a 30-day term, gives you exclusive use of the model, and requires approval: it can take up to 24 hours to validate the contract and enable exclusive access.

For developers

If your code already uses the OpenAI library for chat requests, you change the base address and the key. Chat Completions and streaming are supported. Tools and JSON mode are not available today, so ask for JSON in the prompt. The developer docs list which parameters are used and which are ignored.

For AideaMaker software

Your AideaMaker software can run on Private AI too. A hosted or embedded agent can be set to use AideaMaker Private AI with a key from your own Private AI account, and custom builds can list it as a provider next to the others. You can also stay on your own key from any major provider; it is your choice.

What it does not promise

  • Regulated data. Health records, card numbers and other regulated data need a written agreement with us before they go through Private AI.
  • Capacity. On per-token use, a call can be refused with a retry time when we are at capacity. A rental gives you the model's capacity to yourself for the term.
  • Every OpenAI feature. It covers the chat requests described above, not every OpenAI endpoint.

Where to go next

Create a free account and try it on a real prompt from your own work: start at Private AI services, read the developer docs, or ask Ada which AI your software would run on. If your work involves a regulated setting, book a free consult and we will talk through the written agreement first.