Getting Started

Contextflo works inside the AI you already use. Connect it to Claude, ChatGPT, Cursor or any other MCP client, then ask your agent to "onboard me". It walks you through the rest.

What you need

• An AI agent that supports MCP: Claude, ChatGPT, Cursor, or any other MCP client.
• Your data: read access to a warehouse (Snowflake, PostgreSQL, MySQL, Amazon Redshift, Google BigQuery, Databricks or ClickHouse), or a CSV upload. No data ready yet? Use our demo data.

Step 1: Connect your AI agent

After you sign in you land on Home, and a welcome dialog shows how to connect your agent. You can also open it any time from the Connect your agent button on Home.

Pick your agent's tab and add Contextflo as a custom connector using the MCP server URL shown there. Step-by-step guides:

• Connect Claude
• Connect ChatGPT
• Cursor and other clients: add a custom MCP connector named "Contextflo" with the URL from the Cursor / other tab.

Step 2: Ask your agent to "onboard me"

In a new chat, ask your agent to onboard me. It checks what your workspace already has and walks you through what's missing, one step at a time:

1. Connect your data. It sends you the right connection form for your warehouse and tells you what to fill in.
2. Pick your tables. Choose the tables Contextflo can query. You can change these later on the Data Sources page.
3. Run your first query. Ask a real question about your data and get an answer, with the SQL behind it.

Prefer clicking? The Get to your first answer card on Home has the same steps: connect your AI agent, connect your data, add a context source (optional), and run your first query. Each item ticks itself off as soon as it's done.

Try it with demo data

No database handy? On the setup card, click or try demo data next to "Connect your data". Pick a sample dataset and Contextflo adds it to your workspace with tables, context, metrics and saved queries already in place, so you can start asking questions right away.

Add a context source (optional)

Context sources teach Contextflo what your data means: dbt models, documentation, data dictionaries. Connect GitHub or GitLab, or paste text directly from the Context Sources page. The more context Contextflo has, the more accurate your answers.

Step 3: Make it ready for your team

Once your first query runs, Home switches to Make it ready for your team:

• Add business context: describe what your company does and how to read your data, in Settings > Org info. Every answer gets better for everyone.
• Define your first metric: lock in a definition like "revenue" or "active users" on the Metrics page, so everyone gets the same number.
• Invite a teammate: add people from Settings > Members. They connect their own agent and start asking right away.

Your agent can help with these too. Just ask it after your first query.

What's next

• New user guide: tips for getting accurate answers.
• Dashboards: turn answers into live, shareable dashboards.
• Metrics: define the numbers your team relies on.