The Cloud Pod | Weekly AI & Cloud News on AWS, Azure & GCP · Justin Brodley, Jonathan Baker, Ryan Lucas and Matt Kohn | Cloud Computing & AI News

Databricks One: Because Seven Eight Nine

·1 hr 22 min·5 clips

Dirk Jan Molema found two critical Entra ID vulnerabilities that could have exposed every Azure customer tenant.

1. The Cloud Pod episode on September 23rd, 2025 covers AI, AWS, GCP, and Azure news from one weekly roundtable. 2. Justin, Jonathan, Ryan, and Matthew host the discussion, and their cloud-platform backgrounds shape the practical reactions to each announcement. 3. The episode asks what new cloud releases matter for developers, security teams, and platform operators right now. 4. Google and Kaggle announce a five-day intensive course on AI agents running from November 10th through the 14th. 5. The hosts compare that course with Google's earlier Gen AI course, which attracted 280,000 learners. 6. The curriculum includes agent architectures, tools, memory systems, production deployments, code labs, and a capstone project. 7. The discussion frames AI agents as autonomous systems that can perform tasks, make decisions, and interact with APIs. 8. AWS adds Qwen 3 from Alibaba and DeepSeek V3-1 to Amazon Bedrock as fully managed foundational models. 9. The hosts call out Qwen 3 Coder-480B, its 480 billion total parameters, and its 35 billion active parameters per request. 10. They also note DeepSeek V3-1's thinking and non-thinking modes, code-generation improvements, and support for more than 100 languages. 11. Amazon RDS preview environments now expose MySQL Innovation Release 9.4 with automatic deletion after 60 days. 12. The hosts contrast MySQL Innovation Releases with LTS versions such as MySQL 8 and 8.4, which receive up to eight years of community support. 13. AWS Organizations now supports the full IAM policy language for service control policies, including conditions and not-action elements. 14. The hosts focus on how that change can target specific S3 buckets or EC2 instances across accounts instead of using blanket service restrictions. 15. Amazon Q developer CLI adds remote MCP support so Atlassian and GitHub tools can be managed centrally over HTTP with OAuth. 16. AWS also ships the Nova Act Extension for VS Code, Cursor, and Kiro, letting developers build browser automation agents with natural-language prompts. 17. Google embeds Security Command Center insights into GCE and GKE consoles, surfacing misconfigurations, vulnerabilities, and active threats. 18. Google also expands its MCP database toolbox to Firestore, and the 2025 DORA report shows 90% AI adoption with 80% reporting productivity gains. 19. The episode stays conversational, skeptical, and tool-focused, with the hosts repeatedly weighing convenience against security, pricing, and operational risk. 20. Cloud engineers, security teams, and platform admins will get the most value; casual listeners without cloud tooling interest may skip it.

As heard by us

A cloud-news episode that turns AI hype and identity risk into something operational.

The episode starts with a Databricks joke, then settles into a clear run through AI agent training and Entra ID security. Google and Kaggle's five-day AI-agents course gets treated as a useful teaching moment, with agent design, tools, memory systems, and production deployment…

Read the full review in PlayNext →

Why you'd press play

You get the Databricks dad joke, the five-day agent-course angle, and the Entra ID scare in one pass.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from The Cloud Pod | Weekly AI & Cloud News on AWS, Azure & GCP