In private pilot. Public release planned for 2027.
AI for your whole team, on machines you own.
FlowtheraSI is a suite of private AI tools: a model server, a secure API, a coding assistant, document search, workflows and more. They run on your own server and computers, so there is no per-prompt bill and your data never leaves.
Free tier in every product. No account needed for the desktop app.
On your server
On your computer
On your phone
What changes when AI runs on your own machines
The same kind of help your team gets from cloud AI, without the meter and without sending your data away.
| Topic | Cloud AI subscription | FlowtheraSI |
|---|---|---|
| What you pay | Per prompt or per seat. The bill grows as your team uses it more. | A fixed licence, a free tier, and hardware you own. No token quota. |
| Where your data goes | To the provider’s servers, every time you ask. | Nowhere. It stays on your server and your computers. |
| Which models | The provider’s models, on the provider’s terms. | Open models you choose, with each licence shown before download. |
| Without internet | Stops working. | Keeps working once a model is downloaded. |
| Where it is stronger | The very largest models and the hardest tasks. | Privacy, predictable cost and control over your own tools. |
Every service, and what it is for
On your server
One install on a machine you control. Your team, apps and workflows connect to it.
On your computer
Flowthera Studio, one free app for Windows, macOS and Linux.
- Flowthera StudioThe desktop app: private AI on your Windows, Mac or Linux computer.In development
- Flowthera CodeAn AI coding assistant that never changes your code without your approval.In development
- Flowthera DocumentsAsk questions about your own documents and get answers with the page cited.Planned
- Flowthera ToolboxEveryday developer utilities that work offline, no AI required.In development
- Flowthera Model FinderFind the AI models that will actually run well on your hardware.In development
On your phone
A free companion for Android and iOS.
Built first for three kinds of teams
Software agencies
One server in the office gives every developer a private coding assistant and an API for client projects. Client code never leaves.
See how it works for youHosting and IT providers
Offer your customers a ready, secure AI server on the machines you already run, with updates and backups handled.
Uses Server, Gateway and Mobile
See how it works for youDocument-heavy teams
Ask questions across contracts, reports and invoices, get the page cited, and let routine document work run on its own.
Uses Documents, Automate and Server
See how it works for you
Start with a pilot on your own hardware
We are working with a small number of teams before the public release. Pilot customers get the people building it, pilot pricing, and a say in what comes next.
Request a pilot- 1
Tell us your goal
A short call about what your team wants to do, and the hardware you have or could use.
- 2
We install it with you
Remotely, on a machine you choose. We pick models that fit and set up keys for your team.
- 3
Use it and measure
Your team works with it for a few weeks. We measure setup time, support needed and what it saved.
Private by design, honest about limits
Your data stays put
Prompts, files and answers stay on your machines. Our cloud only handles accounts, licences and updates.
Security and privacyNothing changes without you
Agents propose, people approve. Code edits, commands and workflow steps wait for a yes.
How it worksWe publish the limits
What local models can and cannot do, which systems are supported, and measured results before release.
Our limitations pageQuestions teams ask before a pilot
What can I use today?
Flowthera Server and Flowthera Gateway are in private pilot with early customers. Flowthera Studio and its tools are in development, and the other services are planned. Each page shows its status.
Is AI on our own machines good enough?
For many everyday tasks, yes: drafting, summarising, answering questions about your own code and documents. The very largest cloud models are still stronger on the hardest tasks, and we say so on our limitations page.
Do we need to buy a GPU?
Not to start. Small models run on an ordinary server or a laptop with 8 GB of memory or more. A GPU makes larger models fast, and the model finder shows what fits before you buy anything.
What does it cost to start?
Every service has a free tier. Paid plans start at $19 a month for one person. During the pilot, prices are agreed with each customer.
Can we still use cloud AI models?
Yes, if you choose to. Add your own key for a cloud provider and allow it per team. It is off by default, and every request that leaves your network is labelled.
Who installs and maintains it?
The installer does most of it and updates itself, with automatic rollback. Pilot customers install with us. We also offer a paid setup service.
See it on your own hardware
Tell us what you want to achieve. We will reply within two working days with a suggested setup.