Uploaded April 2026 | Updated September 2026, 2 weeks ago

---

Move beyond prompting and learn to design and scale AI agents with context engineering. In this hands-on session, you’ll control what your agent sees, remembers, and prioritizes for reliable performance in complex workflows.

What You’ll Learn:

- The 5 context types: input, runtime, compression, isolation, long-term memory
- Context compression, middleware, and subagent-based task delegation
- How to prevent common AI failure modes at scale

What You’ll Build:

A research AI agent that compresses context, delegates tasks, enhances tool calls, and manages long-term memory across multi-step workflow

#SambaNova #ContextEngineering #AIagents #AutonomousAI
------------
Table of Contents:

0:00 Introduction
5:11 What is Context Engineering
6:42 Key Problems (Token Limits & Attention Loss)
8:46 Types of Context
13:09 Context Compression Explained
20:27 Compression Results & Impact
23:23 Middleware (Core Concept)
33:33 Sub-Agents & Context Isolation
49:55 Sub-Agent Workflow Demo
54:50 Token Savings & Efficiency
56:52 Key Takeaways