In the cloud or on your own server
Datachat deploys flexibly: as Azure multi-tenant SaaS or on-premises, with automatic data refresh and open integrations.
Technology foundation
Python / FastAPI backend, React / Vite frontend, MSSQL and ClickHouse data warehouse, LiteLLM model abstraction. Answers stream in via Server-Sent Events — no waiting for full processing.
Azure multi-tenant
Each customer gets a dedicated environment — its own Azure App Service, SQL database and ClickHouse instance. Isolation is guaranteed and data never mixes.
Automatic ETL agent
A standalone Windows service syncs the customer’s data into the Datachat warehouse every 5 minutes — invisibly to the user, always with fresh data.
- Incremental sync protected by a safety window
- Automatic detection and addition of new columns
- Admin dashboard with heartbeat, per-table status and error log
- Hourly health-watch alert if the agent goes offline or data goes stale
On-prem and hybrid
For regulated industries, VPN-only on-prem deployment on IIS + Windows Service, optionally Docker Compose. Hybrid LLM: basic mode local, deep analysis in the cloud.
Administration
User and role management, a metadata registry with automatic schema import and relationship detection, data-model export/import in JSON, usage statistics and LLM cost estimates.
MCP integration
Model Context Protocol support for custom tools and data sources — Datachat fits into your existing AI toolchain.
See it work on your own data
We’ll show you a tailored demo built on your Octopus8 data warehouse. 30 minutes, no commitment.