# flurry

> flurry hosts React data apps that an AI agent builds with its MCP tools, for Snowflake. Every app is versioned, and every write follows a declared rule and leaves an audit record.

## Start

For agents: read https://dvt.dev/docs/connect.md first; the snippets there are the exact commands.

- [Start here](https://dvt.dev/docs/index.md): Three ways in, depending on what you do with flurry.
- [Try flurry on your laptop](https://dvt.dev/docs/local.md): Three commands, sample data, no Snowflake account needed.
- [Connect your AI](https://dvt.dev/docs/connect.md): One command or one click per app, then one prompt.
- [Your first app](https://dvt.dev/docs/first-app.md): Four prompts, from a quick look to a write that follows the app's rules.
- [Runbook for AI agents](https://dvt.dev/docs/agents.md): How an agent connects itself to flurry and gets started.
- [How flurry works](https://dvt.dev/docs/how-it-works.md): Versions, rules for writes, the audit trail and trust levels.
- [What leaves your account](https://dvt.dev/docs/what-leaves-your-account.md): Nothing, unless you turn on a connector. Here is exactly what each one sends.
- [Troubleshooting](https://dvt.dev/docs/troubleshooting.md): Each error and its fix.

## Pages

- [Home](https://dvt.dev/): What flurry is and how to ask for early access.
- [Product](https://dvt.dev/product/): The four things flurry does: build React data apps, build them from a sentence, maintain them with versions, and govern every write.
- [Examples](https://dvt.dev/examples/): Example apps running on sample data.
- [Docs](https://dvt.dev/docs/): How to connect your AI to flurry, build a first app, and what to do when something fails.
- [Blog](https://dvt.dev/blog/): Building data apps on Snowflake.
- [Pricing](https://dvt.dev/pricing/): $100/month or $1,000/year, flat for the whole Snowflake account, with a sourced comparison.

## Blog

- [What Snowflake now gives you for building data apps](https://dvt.dev/blog/what-snowflake-gives-you.md): A sourced tour of Cortex Code, App Runtime, Streamlit, artifacts and MCP in Snowflake, and what the docs say about running many apps.
- [The fiftieth app](https://dvt.dev/blog/the-fiftieth-app.md): An agent can now build a Snowflake data app in an afternoon. The first one is easy. The problem starts when a company has fifty.