How to Get Started with Sigma Computing

Sigma Computing connects directly to cloud data warehouses and lets you analyze data using a spreadsheet interface, with no SQL required and no data exports needed. The platform reached $100 million in annual recurring revenue in April 2025, reflecting strong demand from operations teams and analysts who need warehouse-scale analysis without engineering overhead. An April 17, 2026 product release upgraded its AI assistant and added an MCP server for AI tool integrations, making it one of the more fully-featured options available today.

This guide walks through setting up a Sigma account, connecting a data source, and running your first AI-powered analysis.

What Sigma Computing Does Differently

Most business intelligence tools require analysts to either write SQL or wait for a data engineer to build a report. Sigma takes a different approach: it connects directly to your data warehouse (Snowflake, BigQuery, Redshift, or Databricks) and presents the data in a spreadsheet-style interface where formulas, filters, and calculations work the same way they do in Excel or Google Sheets.

The key difference is scale. Sigma queries your warehouse directly, so billion-row tables respond in seconds. No data is copied to a local file and no extracts are created. Every analysis runs live against the source.

Sigma has been named Snowflake's Business Intelligence Data Cloud Product Partner of the Year three years in a row, which reflects how closely it integrates with Snowflake specifically, though it works equally well with BigQuery, Databricks, and Redshift.

Prerequisites

To get started with Sigma, you need a Sigma account (free 14-day trial available at sigmacomputing.com), access to a supported cloud data warehouse, and admin credentials for your warehouse connection. Supported warehouses include Snowflake, BigQuery, Amazon Redshift, Databricks, PostgreSQL, Google Cloud SQL, and Azure Synapse.

Sigma is entirely browser-based, so no software installation is required on your machine.

Step 1: Create Your Account and Organization

Go to sigmacomputing.com and click "Start for free." During signup, you create an organization workspace, which is the shared environment where all workbooks, dashboards, and data connections live. Name it after your team or company.

Once logged in, navigate to Administration (gear icon, top right) to access connection settings, user management, and AI configuration. Only Admin account types can configure data connections and AI features. If you are signing up as an individual, you start as an Admin automatically.

Step 2: Connect Your Data Warehouse

From the Administration panel, click Connections, then Add Connection. Select your warehouse type and enter your connection credentials.

For Snowflake, you will need your account identifier (formatted as accountname.region), warehouse name, database, schema, and username and password or OAuth credentials. For BigQuery, Sigma authenticates via a Google Cloud service account. You provide the project ID and upload the service account JSON key file.

After entering credentials, click Test Connection to verify the setup. A green confirmation message means Sigma can reach your warehouse. Click Save to store the connection.

One common issue with Snowflake: Sigma's IP addresses need to be added to your Snowflake network policy. If the connection test fails, check that Sigma's IP range is allowlisted. Sigma publishes its IP list in the documentation under Connection Requirements.

For BigQuery, connection failures usually mean the service account is missing the BigQuery Data Viewer and BigQuery Job User roles. Grant both roles in Google Cloud IAM before retrying.

Step 3: Create Your First Workbook

A workbook in Sigma is the equivalent of a spreadsheet file. Click New Workbook from the home screen. Inside the workbook, click Add Element and select Table. A data source panel opens on the right side of the screen. Navigate to your connection, drill into your database and schema, and select the table you want to analyze. Click Confirm.

Sigma runs a live query against your warehouse and renders the results in a spreadsheet grid. You can scroll through millions of rows without pagination or lag. Drag columns to reorder them, click a column header to sort or filter, and use the formula bar to write calculations exactly as you would in Excel.

Sigma supports standard spreadsheet functions (SUM, AVERAGE, IF, COUNTIF), date functions, and SQL-style aggregations. If you want to group data, click a column header and select Group Column. Sigma creates an aggregated view automatically.

Step 4: Use the AI Analysis Features

Sigma's AI features require an Admin to enable an AI provider first. Under Administration, go to AI Features and connect either OpenAI or Azure OpenAI using your API key. Once configured, AI features appear throughout the interface.

Sigma Assistant: The April 2026 update renamed Ask Sigma to Sigma Assistant and added enhanced semantic search and AI context for data models. To open it, click the Sigma Assistant button in the top navigation bar. Type a question in plain English, for example: "What were total sales by region last quarter?" Sigma generates a query, runs it live against your warehouse, and returns a chart or table with an explanation of the steps it used. You can click "Open in Workbook" to convert the result to an editable analysis you can build on further.

AI Formulas: In any column, click the formula bar and then select the AI icon. Describe the calculation you want in plain English. Sigma generates a formula and explains what it does. This is useful for complex conditional logic, date calculations, and text parsing that would otherwise require SQL knowledge.

LLM Column Functions: For workbooks that contain unstructured text, Sigma can run LLM-powered functions row by row. For example, you can add a column that classifies customer feedback as positive, negative, or neutral, or extracts specific values from freeform notes. This runs at warehouse scale without requiring a separate Python pipeline or data export.

If your workflow involves raw files rather than a connected warehouse, VSLZ AI lets you upload a CSV directly and run natural language analysis from a single prompt with no infrastructure setup.

Step 5: Build a Dashboard

Once you have workbook elements that you want to track regularly, click Add to Dashboard at the top of any element. Sigma's dashboard editor lets you arrange charts, tables, and text in a grid layout. Each element stays connected to its live data source, so dashboards refresh automatically when underlying data changes.

To share a dashboard, click Share in the top right corner. You can invite users by email, share with your whole organization, or generate an embed link for external stakeholders. Sigma also supports scheduled email exports, where dashboards are sent as PDF or PNG snapshots on a daily or weekly cadence.

Step 6: Connect the MCP Server (Optional)

The April 2026 release added an MCP server that allows AI tools like Claude to query your Sigma organization directly through a chat interface. To configure it, go to your Sigma profile settings, click MCP, and copy the MCP server URL. In Claude or another supported AI assistant, add a custom connector and paste the URL. Authenticate with your Sigma credentials.

Once connected, you can ask questions about your Sigma data and search across workbooks without opening the Sigma UI. This is useful for teams that already use AI assistants for daily work and want data lookups embedded in the same workflow.

Practical Summary

Sigma Computing works well for teams that already have data in a cloud warehouse and want analysts to query it directly without routing every report through a data engineering team. The spreadsheet interface lowers the barrier for non-technical users, the AI formula assistant handles complex calculations without SQL, and the April 2026 Sigma Assistant update makes natural-language querying substantially more capable. Start with the free trial, connect your warehouse, and build a workbook before exploring AI formulas or the MCP integration.