Read receipts and invoices
A vision model reads the supplier, date, amounts and VAT from photos and scans, for your own scripts or any tool that speaks the OpenAI API.
OfflinAI Server puts AI on a machine in your practice. Read receipts and invoices, draft client letters, summarise long documents and transcribe meetings — while every client file stays on hardware you control.
Practical tasks for a busy practice. A qualified person checks the result — AI drafts and reads, you decide.
A vision model reads the supplier, date, amounts and VAT from photos and scans, for your own scripts or any tool that speaks the OpenAI API.
Engagement letters, replies to clients and plain-English explanations of a position — first drafts in seconds, for you to check and send.
Long correspondence, agreements and reports summarised; two versions of a document compared.
Embeddings and rerankers let a document-search application find the right paragraph in your own files.
Client meetings and voice notes turned into text with timestamps, on your own machine.
Content filters mask IBANs, card numbers, email addresses and phone numbers before a model sees them (Business plan).
OfflinAI Server is the AI engine; your team works in the applications that suit them.
OfflinAI Server is not accounting software: it does not keep books or file returns. AI Accountant, our accounting application, is being rebuilt as part of OfflinAI Workspace.
Small models run on a recent laptop. For a team, a desktop or small server with a modern NVIDIA graphics card is the usual choice — and the server also runs on the CPU alone, more slowly.
Tell us your tasks and your team's size, and in a free 20-minute call we will recommend a machine — or tell you whether the one you have will do.
No product can do that on its own. What OfflinAI Server changes is where the data goes: client files are processed on your own machines and OfflinAI never receives them, so no outside AI provider receives client data. Your obligations as data controller stay as they are.
No. OfflinAI Server is an AI server: it reads, writes and summarises on request, through the applications you connect to it. It does not keep books or file returns.
AI models make mistakes, and smaller models make more of them. Treat what they write as a first draft for a qualified person to check — figures above all. A Private AI Pilot tests models on your own three most valuable tasks before you rely on them.
One price per server, not per person. Pro costs €49 a month, €490 a year, or €990 once for a perpetual licence; founding customers get 50% off the first year. The Free plan costs nothing and runs on one machine.
Tell us about your practice and the tasks you have in mind. We'll suggest the hardware and the models, and get you a licence to try it.
Where would you like to run OfflinAI Server, and for what? We will be in touch by email.