🇬🇧 English · Home · Deutsche Version
Sovereign AI for professionals bound by secrecyRuns on any PC. You pay for computing time by the minute.
No inference server, no expensive graphics card, no forced cloud. Every Navigator, from the free Light to Enterprise, runs on a perfectly ordinary PC or laptop. The AI computing power comes from our inference, billed by the minute, on German hardware if you wish, and decrypted only where the computing happens.
€0 · Windows, macOS & Linux · free forever
🧪 Beta: 30 days of fleet:de free for the first 60 · the FD number comes with the download link, the 30 days start the day you enter it
- Runs without inference hardware
- from 50 ct / active minute
- Your data stays on your computer
From a single workstation to a server: four levels
Light, Maat and Enterprise run on ordinary hardware and differ in the number of assistants, the data volumes and the networking, as well as in image analysis, which is reserved for Maat and Enterprise. The Server is the multi-user solution in your own office.
Your personal AI workstation: local, free, permanent.
- ✓30 days of fleet:de free: the FD number comes with the download e-mail, the 30 days start when you enter it in the Navigator
- ✓2 assistants: Helene plus one free slot
- ✓RAG per assistant 5 GB · knowledge base 25 GB
- ✓50 session documents · 50 projects
- ✓Web search, editorial room
- ✓Own or routed inference; connects to Navigators
- ✕No image analysis, images in the chat start with Maat
- ✕Single workstation · no multi-GPU
The lean workstation with no computing load of its own: inference comes from an Enterprise Navigator or via fleet:de.
- ✓Unlimited assistants & data volumes
- ✓No inference of your own needed, runs on any office PC
- ✓Image analysis & OCR: photos, scans and screenshots right in the chat
- ✓Unlimited Mates (browser, e-mail, Office)
- ✓No user limit
- ✓Inference from an on-site Enterprise Navigator or billed by the minute via fleet:de
The full version for several professionals: separate rooms, one machine.
- ✓Everything unlimited: assistants, data, Mates
- ✓Image analysis & OCR: photos, scans and screenshots right in the chat
- ✓Multi-user: a separate, sealed room for each person, no user limit
- ✓Own and routed inference, multi-GPU
- ✓Networks Maats & further Navigators
The multi-user server on your own premises, for law firms, clinics and public authorities.
- ✓Multi-user server on your premises, workstations connect to it
- ✓Central user and permissions management
- ✓Tenant separation & audit logs
- ✓Your own inference hardware on the premises, with no limits
- ✓Optional service & maintenance for €500 / month, backups included
Licence and computing time, cleanly separated
The edition is your software licence. You book the AI computing power separately: billed by the minute, active generation time only, no connection or waiting time.
| Edition | Month | Year | One-off purchase |
|---|---|---|---|
| Light | €0 | €0 | n/a |
| Maat | €89 | €890 | €5,340 |
| Enterprise | €249 | €2,490 | €14,940 |
| Server | €700 | Optional service & maintenance: + €500 / month, backups included | |
| Inference | 60 ct / min with a customer account · 50 ct / min with account and SEPA direct debit · billed by the minute · Beta: the first 60 compute free for 30 days, then €20 starting credit, FD number from the Light download e-mail | ||
You only need a plan if you have no inference hardware of your own
The Navigator computes on your own AI hardware or with fleet:de. Either way, all your data stays on your computer. We sell privacy with AI, not your data.
The first 60 compute free for 30 days. Condition: you actually use it.
fleet:de is live, in beta. We now want real load on the system. Everyone who activates fleet:de in the Navigator during the beta and works with it computes free of charge for 30 days: no base fee, no per-minute billing. No contract that keeps running afterwards. After the 30 days you decide whether a plan fits, and which one. This applies to the first 60 testers. If you come later, the FD number gives you €20 of starting credit for AI minutes. Beta does not mean less protection: GDPR, the data processing agreement and the Section 203 undertaking apply from the first minute, exactly as in regular operation. The contract with terms, data processing agreement and Section 203 commitment is concluded on activation as always.
How to get in: download Light. The FD number comes with the download link by e-mail. Enter it in the Navigator, and from that day 30 days of fleet:de run free of charge. Your usage is shown in the program at the regular price, charged at €0: afterwards you know exactly what fleet:de costs. Questions? info@fleet-data.de.
And the Navigator itself? Costs nothing. Navigator Light is permanently free, for Windows, macOS and Linux. Get Light, enter the FD number, use fleet:de for 30 days: €0 for software and computing time.
Your own hardware
You have a GPU or an inference server in-house? Then the Navigator computes there. No plan, no computing time, no running costs.
fleet:de
No inference hardware? Then only the individual task goes to fleet:de, and the answer comes back. Content is neither stored nor logged. The only thing recorded is the computing time: measured in seconds, billed in 6-second blocks. German servers. As Flex you can cancel monthly, a 12-month term makes it cheaper, and SEPA direct debit cheaper still. And before the first minute runs, the contract is in place: terms and conditions, data processing agreement and Section 203 undertaking, right when you activate.
For trying it out and for occasional users. No subscription, no commitment.
- ✓30 days free for the first 60 (beta), then €20 starting credit: enter the FD number from the download e-mail in the Navigator
- ✓0 included minutes, minute price depending on your customer relationship:
- with account€0.60
- account + SEPA direct debit€0.50
- ✓Any number of users
- ✓Access: customer account (B2B or private) + prepaid, SEPA optional
Solo users, micro-businesses.
- ✓Included minutes per month: 120 on Flex, 170 on Annual
- ✓Any number of users, one shared allowance
- ✓Access: account, SEPA or prepaid
Medical practices, small law firms.
- ✓Included minutes per month: 250 on Flex, 300 on Annual
- ✓Any number of users, one shared allowance
- ✓Priority processing, business features
- ✓Access: account, SEPA or prepaid
Law firms, tax offices, medical practices with a team, SMEs.
- ✓Included minutes per month: 500 on Flex, 550 on Annual
- ✓Any number of users, one shared allowance
- ✓Team management, priority processing, business support
- ✓Access: account, SEPA or prepaid
How billing works
Measured to the second, rounded to 6-second blocks: a block once started counts in full. A request of 8 seconds costs two blocks, that is 12 seconds, not a whole minute. Ten blocks make a minute, so you can work out the block price in your head.
| Base fee / month | Flex | Flex + SEPA | Annual | Annual + SEPA |
|---|---|---|---|---|
| BUSINESS | €39 | €36 | €32.50 | €29.50 |
| PRO | €79 | €74 | €65.83 | €60.83 |
| TEAM | €149 | €139 | €124.17 | €114.17 |
| Plan | € / minute | ct per 6-second block |
|---|---|---|
| BASIS with account | €0.60 | 6.0 ct |
| BASIS with account + SEPA | €0.50 | 5.0 ct |
| BUSINESS Flex · per extra minute | €0.49 | 4.9 ct |
| BUSINESS Annual · per extra minute | €0.39 | 3.9 ct |
| PRO Flex · per extra minute | €0.39 | 3.9 ct |
| PRO Annual · per extra minute | €0.29 | 2.9 ct |
| TEAM Flex · per extra minute | €0.29 | 2.9 ct |
| TEAM Annual · per extra minute | €0.19 | 1.9 ct |
Included minutes are used up first; after that the plan’s minute price applies. Invoice amounts are rounded to the nearest cent.
Which plan fits?
BASIS is the way in without a subscription. With SEPA direct debit the minute price drops from €0.60 to €0.50.
From about 80 minutes a month, BUSINESS Flex is cheaper than BASIS with SEPA, BUSINESS Annual from as little as about 65 minutes, and that plan includes 170 minutes. The base fee of the subscriptions covers the included minutes, and the tiered minute price rewards moving up to the next level.
Flex or Annual? Flex can be cancelled monthly. Stay for a year and you pay for two months less, get 50 extra included minutes per month and pay 10 ct less per minute beyond the allowance. The term is 12 months, invoicing stays monthly, and you can cancel up to 6 weeks before the term ends; otherwise it renews for another year.
SEPA direct debit pays off on every plan. On BASIS it lowers the minute price, on the subscriptions the base fee: BUSINESS €3, PRO €5, TEAM €10 less per month, on Flex and Annual alike.
No user limits: every plan covers any number of users, who share the minute allowance.
ENTERPRISE is available as an individual offer: dedicated capacity, SLA, onboarding, API and system integration, your own AI workflows. No list price.
When a contract comes into being – and when none is needed
Fleet Navigator draws a clean line: as long as the inference runs on your computer, we process nothing. And where we do the computing, the contract is in place before the first minute runs.
Light, local: no contract needed
If the inference runs on your own hardware, we process no data – no data processing agreement, no undertaking, no involvement on our part. Legally as clean as Word on your computer: download it and get started. Light remains the free way in, with no contractual burden whatsoever.
fleet:de: contract first, then inference
When you activate fleet:de, you accept the terms and conditions, the data processing agreement (DPA) and the Section 203 undertaking – electronically, with a click. Without this consent our server accepts no request. The data processing agreement is concluded with your customer account; business accounts (B2B) conclude it automatically as part of the account. That is why fleet:de is only available with an account: a data processing agreement with a mere e-mail address would not be one – the controller is the law firm, not an e-mail address.
You can see where the computing happens
The Navigator shows where the computing happens as a status at all times: “Inference: local” or “Inference: fleet:de (DPA in place)”. Not a warning, a status display – and no switch ever happens unnoticed. Which cloud chatbot tells you where your request is being computed right now?
Your data stays on your computer.
Everything stays local
Documents, knowledge base and chats never leave your computer, and nothing is uploaded. For the computing, only the current request goes out, encrypted; none of it is stored.
Section 203 as the foundation
Built for doctors, lawyers and everyone whose professional secrecy is laid down in law: Section 203 of the Criminal Code (StGB) in Germany, Section 121 StGB in Austria, Art. 321 of the Criminal Code in Switzerland. We compute for you; we do not hold other people’s files.
Encrypted all the way to where it is computed
One protocol everywhere, end to end. Decryption happens only where the computing happens. On the German route, that is hardware under our control. The route is visible in the client.
No lock-in
If a licence expires, the Navigator falls back to Light, and your data is never locked in. You are free to leave at any time.
With us, full data sovereignty is not an advertising line but part of the contract.
Fleet Data & System Consulting UG guarantees you by contract:
- ✓Encrypted twice. Transport encryption plus end-to-end encryption of every request. Decryption happens only on the compute node itself.
- ✓No outside access. Nobody but us has access to our inference machines: no cloud provider, no service provider, no third party.
- ✓No logging. Requests and answers are neither logged nor stored. Every inference request and its answer are seen only by the user.
- ✓Section 203 StGB and GDPR. For every piece of confidential information that matters to you: fit for Section 203 and the GDPR, guaranteed by contract.
In short: your data is processed as if in a safe deposit box to which only you hold the key. Just not at the Sparkasse savings bank in Gelsenkirchen, where burglars emptied around 3,200 safe deposit boxes in December 2025. No underground car park leads to our machines.
And if you would rather not rely on any promise at all: buy your own inference hardware, an Apple Mac or an Nvidia graphics card. What fits is shown in our hardware overview. There, physics rules. And physics beats promises.