Credential reset
Change local passwords and admin credentials after delivery so your organization controls access.
Security
An inspectable local AI appliance you own.
A SelbsAI Server belongs to you, runs models on its own dedicated compute and memory, and does not need a remote model API for normal AI work. The security promise is practical and visible: immutable system partition, no hidden agents, no document cloud sync, and no training on your files.
Dispatch integrity
The dispatch image is verified before release.
Every SelbsAI Server and AI Boot Drive release is scanned after provisioning and again before packaging. A detection, unexpected executable, or checksum mismatch quarantines the release automatically: it does not ship.
Verified release image
Verification covers the software image and files supplied by SelbsAI at dispatch. Software, files, models, or updates added after handover remain under the customer’s control.
- 01
Controlled sources
Runtime and model artifacts are obtained from recorded project or publisher sources.
- 02
Hash verification
Recorded checksums must match the exact artifacts installed on the server.
- 03
Full-drive scan
The completed image receives a malware scan before the package can be released.
- 04
Recorded release
Scan result, tool version, signatures, timestamp, image hash, and operator are retained with the order.
Customer-owned server appliance
You receive a complete configured computer and local software stack, not a managed black box. It can be reviewed by your IT contact, updated under your policy, and used without giving SelbsAI continuous access to your documents.
Network choice
Keep it offline, place it on an internal network, or connect it under your normal IT policy.
Update timing
Apply operating-system, runtime, and model updates when your risk and maintenance policy allows.
Clear documentation
Receive setup notes, installed-component visibility, provenance references, and support contacts in the customer portal.
Exactly what we install
The stack is limited to the local AI experience and the tooling needed to operate it cleanly.
Local AI runtime
local runtimeThe inference runtime that runs selected open-weight models directly on the appliance.
Selected model files
open-weight modelsModel families matched to the chosen workload, appliance memory, language needs, and privacy expectations.
Private chat interface
local UIA local browser interface for asking questions, drafting text, reviewing files, and controlling workflows.
Document retrieval tools
local RAGLocal indexing and retrieval components so approved files can be searched without sending them to a model cloud.
Document parsing add-ons
optional parsingOCR, text extraction, and file readers where selected for invoices, contracts, scans, policies, or technical documents.
Role-based presets
use-case setupWorkflows, prompts, and configuration for the selected use cases, such as legal review, advisory work, coding, compliance, or operations.
What we do not install
Security also comes from restraint. The server is not shipped with hidden monitoring, tracking, or background cloud document sync.
No spyware or adware
No tracking toolbars, adware bundles, behavioral advertising SDKs, or consumer data brokers.
No hidden remote-control agent
No permanent selbsai support account or unattended remote access unless a customer separately requests support.
No cloud document sync by default
Your files are not silently mirrored to selbsai, model vendors, or third-party storage for normal AI use.
No training on customer files
Customer documents are not used to train public models or shared model services.
Customer controls after delivery
At handover, the customer can inspect the setup, change credentials, decide update timing, export the workspace, and keep operational control of the appliance.
Owner
Customer-owned server appliance
Boundary
Runs on your premises
AI use
Local inference by default
Access
No standing selbsai admin access