Private conversation
Draft, summarise, translate, compare, and reason through work in a familiar chat interface.
Example · Draft a client letter from these notes
How it works
Your own AI. One local address.
Using selbsai is deliberately familiar. Connect the preconfigured local AI server to power and your approved network, switch it on, then open selbsai.local in a browser. Existing computers stay ordinary clients while the server performs the AI work.
Core AI runs locallyNo model-provider account
selbsai.locallocal
SelbsAI Server
01
Local network
02
Any local browser
03
Private AI ready● ready
The physical product
A local AI server you can point to—not another invisible subscription.
SelbsAI Server is a dedicated x86-64 server appliance in a compact mini-PC form factor. It arrives with memory, internal storage, Linux, the selected model runtime, a private browser workspace and a documented recovery path already prepared.
- Dedicated local compute
- Installed DDR5 + internal NVMe
- Power supply + Ethernet cable
- Provisioned model and runtime
- Recovery and provenance record
Two local access paths
Browser workspace
Open selbsai.local from approved office devices.
SelbsAI Server
● Inference remains on the appliance
OpenAI-compatible API
Connect approved IDEs and agents to the local endpoint.
Local Ethernet / approved LAN
First start
From the box to your first answer.
There are four actions: connect power, connect Ethernet, switch on the appliance, and open the local workspace.
- 01Connect
Connect the SelbsAI Server.
Attach the included power supply and connect Ethernet to the approved local network.
- 02Start
Switch on selbsai.
Press the appliance power button and wait for the ready status shown on the quick-start card.
- 03Open
Visit selbsai.local.
Open the address in Chrome, Safari, Edge, or Firefox from an approved computer on the same network.
- 04Work
Ask, attach, continue.
Choose a workspace, type your request, or add a document. The selected local model handles the task.
The included setup card lists the local address, status check, shutdown, and recovery procedure. No changes to client computers are required.
The everyday experience
After startup, it works like the AI you already know.
Open a browser, choose what you want to do, and write naturally. There is no terminal, command syntax, or model dashboard in the normal workflow.
- Start or continue a conversation
- Attach an approved document
- Choose a focused workspace
- Open it from another device on the same network
http://selbsai.local
selbsai
● local · ready
What would you like to work on?
Client documentsResearch desk
Ask a question or attach a document…
What you can do
Three simple ways to use it.
Document workspace
Work with approved PDF, Word, spreadsheet, and text files without sending the core task to a model cloud.
Example · Find every renewal clause in this folder
Local team access
Open the same interface from a laptop or tablet on the approved local network while the SelbsAI Server performs inference.
Example · Open selbsai.local from an office laptop
The next working day
Start the server. Open your bookmark. Continue working.
The server can remain on in the office. Open your saved selbsai.local bookmark from an approved computer and continue with the local workspaces and conversations stored on it.
01
Start the server
02
selbsai.local
03
Continue working
Where the work goes
Your prompt takes the short route.
The browser sends work across your local network to the SelbsAI Server. Its own memory and compute run the model and return the answer locally.
You
local
Browser
local
Local network
local
SelbsAI Server
local
Model cloudNot in the core path
Checkout, optional updates, support, and optional preload use clearly defined online services. Core local inference does not require a model-cloud account.
Product scope
What is included now—and what is not.
Included with every server
- Dedicated x86-64 compute, DDR5 memory and NVMe storage
- Preconfigured Linux runtime and curated open-weight model
- Private browser workspace at selbsai.local
- Power supply, Ethernet cable and setup card
- Recovery media and model-provenance record
Optional services
- Pre-Load document indexing
- Additional workflow presets
- Guided network and team setup
Not sold as available today
- Automatic network-folder synchronisation
- One-click encrypted workspace export
- Fully automatic model upgrades
Choose capacity, not complexity
The workflow is the same on both servers.
Core and Pro open at the same local address and use the same browser workflow. The difference is included memory, storage, and the model and concurrency headroom inside the appliance.
8 GB DDR5 · 1 TB NVMe
Sai Start Server
€499.00 net
€593.81 incl. DE VAT
excl. VAT · prepared to order · 5–14 business days
Best for
Private drafting, document Q&A, transcription, research, and focused office workflows.
- Included memory
- 8 GB DDR5
- Model range
- 7B–14B quantized
- Included with both
- Power supply, Ethernet cable, setup card
16 GB DDR5 · 1 TB NVMe
Sai Personal Server
€699.00 net
€831.81 incl. DE VAT
excl. VAT · prepared to order · 5–14 business days
Best for
Larger document collections, coding, multi-workflow setups, and high-memory local inference.
- Included memory
- 16 GB DDR5
- Model range
- 14B–32B quantized
- Included with both
- Power supply, Ethernet cable, setup card
32 GB DDR5 · 1 TB NVMe
Sai Core Server
€899.00 net
€1,069.81 incl. DE VAT
excl. VAT · prepared to order · 5–14 business days
Best for
Private drafting, document Q&A, transcription, research, and focused office workflows.
- Included memory
- 32 GB DDR5
- Model range
- 7B–14B quantized models
- Included with both
- Power supply, Ethernet cable, setup card
64 GB DDR5 · 2 TB NVMe
Sai Pro Server
€1,499.00 net
€1,783.81 incl. DE VAT
excl. VAT · prepared to order · 5–14 business days
Best for
Larger document collections, coding, multi-workflow setups, and high-memory local inference.
- Included memory
- 64 GB DDR5
- Model range
- 14B–32B quantized models
- Included with both
- Power supply, Ethernet cable, setup card
If you can use a browser, you can use selbsai.
The appliance handles setup and inference. Everything your team uses day to day stays in a familiar browser.