--- title: "Workshop: 300 Right Tokens Beat 100k — Fix Your AI Agent Context" description: Most AI agents fail because of bad context, not bad models. Learn the 4 context antipatterns killing your agent's performance and how to fix them. image: https://info.perfectscale.io/hubfs/Four%20Context%20Anti-Patterns%20Killing%20Your%20AI%20Agent%20%E2%80%93%20Light%20%E2%80%93%201.png --- ![PerfectScale by doit logo](https://info.perfectscale.io/hs-fs/hubfs/PerfectScale%20by%20doit%20logo.png?width=250&height=61&name=PerfectScale%20by%20doit%20logo.png "PerfectScale by doit logo") ## HANDS-ON OPTIMIZATION WORKSHOP  # The Right 300 Tokens Beat 100k Noisy Ones  ### Four Context Antipatterns Killing Your AI Tuesday, June 9, 2026, 11:00am ET Your agent has 100k tokens of context. It still forgets what you told it two messages ago. Join Baruch Sadogursky and Anton Weiss for a live session on context engineering, the architecture decision most teams skip. Context isn't just a prompt setting. It's a design surface that you can get wrong in four very specific ways. This session dissects each antipattern with a live coding agent demo: what breaks, why it breaks, and what the fix actually looks like in practice. Spoiler: it's not about throwing more tokens at the problem. The teams burning the least context are running the tightest architectures. What You'll Learn  - The four context antipatterns: Stuffed Prompt, Wrong Tool for the Job, Goldfish Agent, and Vibes Eval - A decision framework for context architecture that doesn't require a PhD in prompt engineering - Four checks you can run on your own agent Monday morning - Live demo: see each fix applied to a real coding agent **Who Should Attend**  - CTOs and engineering leaders scaling AI-driven infrastructure - DevOps & SRE teams managing Kubernetes in production - Developers building or running AI agents in their daily workflow --- Meet Your Instructors ![baruch no background](https://info.perfectscale.io/hs-fs/hubfs/baruch%20no%20background.png?width=150&height=150&name=baruch%20no%20background.png "baruch no background") **Baruch Sadogursky** Member of DevRel Staff, Tessl AI Baruch has spent years at the intersection of developer tooling and platform engineering. At Tessl, he focuses on how AI agents actually get built and deployed in production. ![Anton Weiss](https://info.perfectscale.io/hs-fs/hubfs/image%20252%20(1).png?width=150&height=150&name=image%20252%20(1).png "Anton Weiss") Anton Weiss Chief Agentic Cluster  Whisperer, PerfectScale by DoiT Anton works directly with engineering teams running AI agents in Kubernetes environments, where context waste has a real compute cost attached to it. [Save Your Seat](https://info.perfectscale.io/workshop-300-right-tokens-beat-100k-fix-your-ai-agent-context#register) PerfectScale by DoiT, 2026