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

338: T5Gemma Says "AI’ll be Back”

·1 hr 2 min·4 clips

Google’s UCP lets AI agents shop across discovery, checkout, and post-purchase support.

1. The Cloud Pod Episode 338, recorded January 13, 2026, centers on AWS, Azure, and GCP news with a strong AI focus. 2. Justin, Jonathan, Ryan, and Matthew host the show, and their roles matter because they trade examples from work, home, and cloud operations while reacting to the headlines. 3. The episode asks what changes when Anthropic, Google, Snowflake, and Flexera push AI into office work, commerce, observability, and cost control. 4. Anthropic’s Cowork feature extends Claude’s macOS desktop app to general office tasks by granting folder access and using plain-language instructions. 5. The hosts mention receipt-photo expense reports, reports from notes, and file recognition as examples of Cowork’s file automation. 6. They compare Cowork with Cloud Code and model context protocol integrations, then note that new requests can be submitted before an initial task finishes. 7. The hosts also discuss trying to use AI for a PowerPoint presentation, including Copilot pulling data from multiple sources and producing imperfect slides. 8. Google’s VO 3.1 update in the Gemini API and Google AI Studio adds enhanced video generation with character identity and background consistency. 9. The model now supports vertical 9:16 output, 4K resolution, improved 1080p quality, and SynthID watermarking. 10. The hosts warn that three-second VO clips can still produce a $35 bill, making cost a practical concern. 11. Snowflake’s acquisition of Observe brings logs, metrics, and traces into the Snowflake platform alongside business data. 12. The hosts question whether Snowflake can replace Datadog, Splunk, or New Relic, and they point out that observability data is expensive to retain. 13. Flexera’s purchases of ProsperOps and Chaos Genius expand its FinOps approach into automated commitment management and data workload optimization. 14. ProsperOps is described as managing more than $6 billion in annual cloud usage and generating over $3 billion in lifetime savings for customers. 15. The hosts debate whether an AI should be allowed to buy savings plans and reservations without stronger approval workflows and guardrails. 16. Google’s Gemini CLI dashboard tracks monthly active users, token consumption, and code changes through Google Cloud Monitoring and OpenTelemetry. 17. The hosts like the usage view because it shows token consumption per hour, but they also joke that there is no metric for how stupid a question is. 18. NotebookLM’s new data tables feature automatically turns scattered source material into structured tables that can be exported to Google Sheets. 19. The episode feels conversational and workmanlike, with the hosts using first-hand examples, jokes, and practical cloud-ops commentary instead of formal interview segments. 20. People who follow AWS, GCP, Azure, AI tooling, and FinOps trends would get the most value, while listeners seeking a narrative story or deep single-topic analysis might skip it.

As heard by us

A practical panel discussion about AI entering spreadsheets, security data, and everyday office workflows.

The episode frames Anthropic's Cowork as a test case for AI-assisted office work inside familiar company files. The panel explains how the macOS Claude desktop app can be given access to selected folders, then used with plain-language instructions for jobs like filling expense…

Read the full review in PlayNext →

Why you'd press play

Want to know what happens when AI wanders into your spreadsheets and vulnerability lists?

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