Nexatron Documentation
What do you want to build?
6 starting guides
Embed in React
Add a governed chat experience to your application.
Connect Claude over MCP
Authorize tools and ask your first analytical question.
Use the JavaScript SDK
Install the SDK and send your first query.
Call from Python
Authenticate and query through the REST API.
Explore the API
Find authentication, requests, responses, and limits.
Evaluate answer quality
Understand corpora, grading, refusal, and release thresholds.
Nexatron, enterprise AI analytics with MCP built in. The goal is a governed answer engine for all of your enterprise data (warehouses, business applications, and documents) reached through natural language and accessible to AI agents over the Model Context Protocol (MCP). It is vendor-neutral and auditable: you bring your own LLM, your data stays where it lives, and every answer is traceable.
Key Features
- Natural Language Queries: ask questions in your own words. Nexatron checks each query before it runs, and it runs as the person asking, so each source's own grants apply. A second, independent check then tests the result. You get structured results with visualizations, alongside the query, the sources, and a trust label.
- Approved-Metric Semantic Layer: define business metrics, dimensions, and relationships once. Every query uses the approved definition, in the right SQL dialect for each source.
- Warehouse and Application Connectors: connect PostgreSQL, MySQL, SQL Server, Snowflake, BigQuery, Databricks, Redshift, Amazon S3 and DuckDB, plus Google Sheets and Microsoft Excel as file sources. Business applications such as Salesforce and QuickBooks connect too.
- Embeddable SDK: drop a chat widget into any React app with a few lines of code.
- MCP Integration: use Nexatron from Claude, ChatGPT, Microsoft 365 Copilot or any MCP-compatible AI assistant through the hosted MCP server. Nexatron acts as both an MCP server and an MCP client.
- Models Where You Run: Claude through the Anthropic API on SaaS, or through Amazon Bedrock in your own AWS account.
- Enterprise Security: multi-tenant isolation with four-layer row-level security, a cryptographically signed audit log, SSO, and API key management.
Quick Links
| Goal | Guide |
|---|---|
| Run a pilot on Snowflake, with no identity-provider work | Pilot quick start |
| Learn the app: Ask, Dashboards, Insights, Knowledge | Using Nexatron |
| Set up and run a workspace | Setup, Sources, Workspace settings |
| Install the SDK and send your first query | SDK Quickstart |
| Embed a chat widget in React | React Widget |
| Connect Nexatron to Claude | Connect to Claude |
| Connect ChatGPT or Microsoft 365 Copilot | ChatGPT, Copilot |
| Use any MCP client, or the local stdio package | MCP Integration |
| Ask from Slack or Microsoft Teams | Slack, Teams |
| Call the REST API from Python | Python Quickstart |
| Browse the public API | API Reference |
Need a custom plan or enterprise features?
Contact your account team or reach out via the in-product help menu for SCIM provisioning, custom rate limits, dedicated infrastructure, and on-prem deployment options.