Agentic AI, Sora 2, LLMs & Model Context Protocol

Send us Fan MailThis week on Tech 4 Thought, we kept it casual and zoomed in on four big themes: agentic AI that actually works in the wild, Sora & Sora 2 updates, where LLMs are headed next, and why the Model Context Protocol (MCP) keeps popping up in dev roadmaps.📚 Topics Covered:• Agentic AI IRL: plan → act → reflect → retry; when short chains beat long ones; tool use that actually sticks• Sora 2: what’s new, creative workflows, and production pitfalls (latency, consistency, control)• LLMs: routing vs. one-model, context windows, evals, and what “good enough” means for teams• MCP (Model Context Protocol): portable tools/data, lighter lock-in, and easier app distribution• Safety & Governance: approvals, audit logs, and cost caps for agent workflows💬 Reactions, takeaways, and hot takes from the whole Tech 4 Thought crew—plus your comments on where agentic AI and Sora 2 are actually useful.🔗 Show notes & socials: https://linktr.ee/tech4thoughtSupport the show