Google Cloud Live: Getting started with MCP Toolbox for Databases

Published on 5 May 2026, 16:46

MCP Toolbox GitHub → goo.gle/github-mcp-toolbox
MCP Toolbox for Databases (Docs) → goo.gle/mcp-toolbox-dev
Blogs → goo.gle/meduim-mcp-toolbox

Ready to bridge the gap between your LLMs and your data? Model Context Protocol (MCP) is changing how we build AI agents. On our next livestream, we’re doing a deep dive into the MCP Toolbox for Databases. Join Kurtis Van Gent and Stephanie Wong to learn how to give your agents secure, high-performance access to AlloyDB and BigQuery.

We’re covering:
* The MCP Toolbox: How to support generic, runtime, and custom tools.
* Secure Data Access: Authorization and authenticated parameters to keep your data safe.
* Real world use cases: A live look at the "Cymbal Flowers" AI agent demo.
* Live Q&A: Get your questions answered by the experts building these tools.

Stop guessing how to connect your data to your agents and start building.

This livestream originally aired on May 5, 2026 at 9:00 A.M. PDT / 12:00 P.M. EDT.

Chapters:
0:00 - Countdown
1:52 - Intro
3:15 - Securing data access in the era of AI agents
4:40 - Modern challenges with securing data and AI agents
9:22 - MCP Toolbox for Databases (Toolbox)
10:54 - Build time agents and run time agents
13:16 - How does MCP Toolbox for Databases stop confused deputy attacks when an AI agent is talking to live databases?
17:00 - Architecting safe and robust AI agents in modern apps
19:11 - [Demo] Cymbal Air
23:44 - [Demo] Trying to compromise the AI agent
26:31 - Demo architecture & code
30:02 - ​​What are the latency trade offs when using MCP vs direct database queries in high throughput systems?
32:38 - Is this a self hosted toolbox or is it Google managed?
33:32 - Does Toolbox support progressive exposure of tools?
35:48 - Can Toolbox write queries on its own?
37:30 - Can Toolbox support complex queries like table joins or stored procedures?
38:09 - What’s the minimum architecture to deploy MCP in a production ready environment?
40:35 - Kurtis final thoughts
42:14 - Wrap up and getting started

More resources:
* GEAR → goo.gle/GEAR
* QuickStart → goo.gle/mcp-quickstart
* MCP Toolbox for Databases:Making BigQuery datasets available to MCP clients (Codelab) → goo.gle/codelabs
* Build a Multi-agent App with MCP Toolbox for AlloyDB & ADK (Codelab) → goo.gle/codelab-multi-agent-ap...
* Cymbal Air Toolbox Demo → goo.gle/4tfWYIA
* Google Cloud MCP servers overview → goo.gle/42ioQRn
* MCP Toolbox for Databases (Toolbox) → goo.gle/4wauUJp

🔔 Subscribe to Google Cloud Tech → goo.gle/GoogleCloudTech

Speakers: Stephanie Wong, Kurtis Van Gent
Products Mentioned: Gemini, Gemini Enterprise Agent Platform, MCP Toolbox for Databases, Vertex AI, Cloud Run