On-prem · Data sovereignty · KVKK/GDPR compliant

Enterprise AI, without your data ever leaving your servers.

Sentinel Mind 360 connects your documents and databases to an AI that runs on your own servers. Answers come only from your data, and each one shows the source it relies on. When it does not know, it does not make things up — if the answer is not in your sources, it says so plainly.

Open Live Demo Explore capabilities

0 data egress 100% source citation Documents + databases

Sentinel Mind 360Enterprise Assistant on-prem

What is the supply voltage and stock level of the XYZ-2200 sensor?

Supply voltage 3.0–5.5 V, current stock 1,240 units, unit price $2.80.

datasheet.pdf · p.2stock · live DB

Fully on-prem KVKK / GDPR compliant Sourced & hallucination-free Homegrown · Turkish & English support

Capabilities

Turn your organization's knowledge into working intelligence

Talk to your documents and databases; every answer you get shows the source it relies on, right beside it.

Document chat

Upload PDF, Word, Excel and CSV files, with OCR for scanned documents. Ask in natural language and get answers with their source.

Database connectivity

Connect SQL Server and PostgreSQL read-only; structured facts such as stock and price come from a live query.

Sourced & hallucination-free

Every answer shows which document or row it draws on. If it is not in the source, it does not invent — it holds back.

Studio outputs

Generate summaries, FAQs, tables, quizzes and mind maps from your documents. All grounded in your documents and delivered with their source.

Agent mode

Multi-step, tool-calling assistants. Actions require human approval; document content cannot trigger an action.

Role & row-level access

RBAC everywhere. An unauthorized user cannot see HR/Finance content — it never even reaches their answer.

How it works

Four steps from install to answer

Install

Containers deploy to your own server. No internet access required.

Connect

Upload your documents and connect your databases read-only.

Ask

Ask in natural language; the system answers only from your data.

Trust

Every answer comes with its source, and every access is in the audit trail.

Data sovereignty

Every component stays within your boundary

The AI model, search engine, data store and interface all run inside your own network. You close off any outbound path — and prove it stays closed.

Data never leavesDefault (Sovereign mode): no document, query or credential leaves the customer network. Open mode engages only if deliberately enabled, and only for non-sensitive tasks.

Tamper-evident auditAll access is recorded in a hash-chained, immutable audit log.

AES-256 & TLSData at rest with AES-256, in transit with TLS 1.2+; passwords hashed.

Human-approved actionsHigh-risk operations require dual approval; a document instruction cannot trigger an action.

Customer network · on-prem

InterfaceBlazor

EngineAPI

LLMon-prem

VectorSQL / pgvector

Auditaudit

Egress to cloud LLM: closed Optional Open mode only for non-sensitive tasks, audited and flagged.

Use cases

One platform for every department

Sales / Technical

Product & datasheet assistant

Find the right technical value across thousands of datasheets in seconds, with its source.

Operations

Stock & price lookup

Structured facts from the live database; up-to-date stock and price, not guesswork.

Human Resources

Policy & procedure

Permission-aware answers over internal regulations; unauthorized content stays invisible.

Support

Technical support helper

Fast, sourced support answers over manuals and resolution documents.

Legal / Compliance

Contracts & regulations

Search across contract and regulatory texts and get summaries with their source.

Management

Enterprise knowledge base

Unify scattered corporate documents into a single, secure and auditable intelligence.

Live demo

See how it works

Click one of the example questions below; see how the answer is formed and which source it relies on. If you like it, try the working version yourself.

Open Live Demo →

Demo login credentials

Administrator (full)demoDemo!2026

Sales rolemehmetParola!123

HR rolezeynepParola!123

The accounts above are role-based, read-only previews. To upload your own documents and experiment freely, open a private temporary workspace with "Start Demo" on the login screen (it resets at the end of the session).

sentinelmind360-dev · /login Sentinel Mind 360

Hello. I answer over your documents and databases, always with the source. Pick an example question on the left.

Sentinel product family

Monitor your data, think with your data

Existing product · sentineldb360.com ↗

Homegrown database monitoring platform. Monitor SQL Server, PostgreSQL, MySQL and MongoDB in real time from a single panel.

New · this product

On-prem enterprise AI. Sourced, hallucination-free answers over your documents and databases, without data leaving your network.

FAQ

Frequently asked questions

The most common questions about on-prem enterprise AI.

Does our data really never leave?

It does not. The AI model, search and interface run on your own servers; your documents, queries and database contents never leave your network. You can close off any egress to a cloud provider — and verify that it stays closed.

Does it work without internet, in a closed network?

Yes. It can be deployed to fully offline (air-gapped) environments. Even license validation is offline; the system does not connect out or "phone home".

Which document and data sources does it support?

PDF, Word, Excel, CSV and text documents; OCR for scanned paperwork. For structured data you connect SQL Server and PostgreSQL read-only. Information such as stock and price comes from a live query, not from the model's memory.

Is there a risk of wrong or fabricated answers?

The system produces answers only from your own sources and shows the document or row each answer relies on. If the information is not in the source, it does not invent it; it plainly says it does not know.

Is it suitable for KVKK and GDPR?

Data stays on your own servers, within your boundary. All access is kept in an immutable (hash-chained) audit log. With roles and row-level access control, an unauthorized user cannot even see restricted content in an answer.

Does it integrate with our existing users and roles?

Yes. Role-based access control (RBAC) comes built in; group-to-role mapping can be connected to your corporate identity provider (LDAP/OIDC). A Super Admin can design your own roles.

Is a GPU required, or does it run on CPU too?

A GPU is not required. It runs on CPU alone on a mid-range server; small and mid-sized models give useful answers on CPU. For organizations that need high concurrency and real-time speed, a GPU (e.g. NVIDIA) is recommended but not mandatory. We size the hardware together based on your load and speed expectations.

Which operating systems does it run on?

On the server side it runs on Linux (Ubuntu, RHEL/Rocky, etc.) with a Docker/container runtime; it can also be deployed on Windows Server via a container environment. On the user side no desktop installation is needed — a standard web browser is enough.

What is the data/DB layer, which database does it use?

The system's own data (users, roles, documents, audit and vectors) is kept on SQL Server; for vector search you can use SQL Server's native VECTOR support or pgvector. You can also connect your existing SQL Server and PostgreSQL databases read-only and include their contents in search. Raw data, embeddings and models never leave your network.

How is it installed, is the source code shared?

Only signed container images are delivered to the customer; source code and build scripts are not shared. It installs on your own server, and the DMC team assists with installation and training.

Build your enterprise intelligence on your own servers

Harness the power of AI without sharing your data with anyone. Take a look at the live demo or send us whatever is on your mind.

Open Live Demo →