Getting more out of the Claude Platform

ℹ️ EPISODE INFO

Released

May 7, 2026

Duration

28m

Channel

Claude

Watch on YouTube

▶ Open ↗

📝 EPISODE DESCRIPTION

Cut cost, manage context, boost intelligence. In this session, we'll show you how to put our latest platform capabilities to work. Through live demos you'll see what great prompt caching looks like, learn to keep context lean for long-running agents with tool search, programmatic tool calling, and compaction, and use the advisor strategy for a cost-effective intelligence boost. Together, they're a set of patterns you can apply to your agents today to get more from every token.

🧾 EPISODE SUMMARY

In this episode of Claude, Getting more out of the Claude Platform explores Production agent performance: prompt caching, context engineering, and advisor models Prompt caching is presented as the highest-impact optimization, delivering major cost reductions, faster time-to-first-token, and relief from API rate-limit pressure for repeated prompt segments.

🔗 RELATED EPISODES

Giving coding agents their own computers: How Cursor built cloud agents

Memory and dreaming for self-learning agents

Getting started with Claude Cowork

How Datadog built a universal machine tool for Claude Code

Architecting for model step-changes: A fireside with Vercel's Guillermo Rauch

Building AI-native: Inside the stacks powering Cognition, Gamma, and Harvey