BigData, Machine-learning, AI & AnalyticsConference50min
Never Trust a Monkey: From AI Slop to Code You Can Ship
This talk introduces a framework for AI-assisted development that bridges the gap between human intent and executing code using three principles—Chasm, Context, and Chain. Drawing from real failures, it shows how to preserve meaning through structured, verifiable workflows, ensuring humans define intent while AI handles implementation.
talk.summaryAiDisclaimer
Baruch SadogurskyTessl AI
We're in the middle of another leap in abstraction.
Like compilers, cloud, and containers before it, AI coding agents arrived with hype, fear, and broken assumptions. We gave the monkeys GPUs. Sometimes they
output Shakespeare. Other times, they confidently ship code that compiles, passes tests, and still does the wrong thing.
The problem is the gap between what we mean and what actually runs.
This talk delivers a practical framework for working with AI agents, built on three ideas: the Chasm between human intent and the code that actually runs,
the Context that replaces guessing with grounding (APIs, conventions, constraints, domain rules), and the Chain that keeps intent alive through a structured
flow from prompt to spec to test to code, where every step produces a verifiable artifact validated externally.
The framework comes from real failure patterns: systems that passed every test, shipped successfully, and still failed to meet intent. Through interactive
demonstrations and honest war stories, Baruch will trace how intent gets lost and build the guardrails that prevent it.
You'll leave with a working model for AI-assisted development where humans own the meaning and machines do the typing.
Trust your context. Trust your guardrails. Never trust a monkey.
Like compilers, cloud, and containers before it, AI coding agents arrived with hype, fear, and broken assumptions. We gave the monkeys GPUs. Sometimes they
output Shakespeare. Other times, they confidently ship code that compiles, passes tests, and still does the wrong thing.
The problem is the gap between what we mean and what actually runs.
This talk delivers a practical framework for working with AI agents, built on three ideas: the Chasm between human intent and the code that actually runs,
the Context that replaces guessing with grounding (APIs, conventions, constraints, domain rules), and the Chain that keeps intent alive through a structured
flow from prompt to spec to test to code, where every step produces a verifiable artifact validated externally.
The framework comes from real failure patterns: systems that passed every test, shipped successfully, and still failed to meet intent. Through interactive
demonstrations and honest war stories, Baruch will trace how intent gets lost and build the guardrails that prevent it.
You'll leave with a working model for AI-assisted development where humans own the meaning and machines do the typing.
Trust your context. Trust your guardrails. Never trust a monkey.
Baruch Sadogursky
Baruch Sadogursky (@jbaruch) did Java before it had generics, DevOps before there was Docker, and DevRel before it had a name. He built DevRel at JFrog from a ten-person company through IPO, co-authored "Liquid Software" and "DevOps Tools for Java Developers," and is a Java Champion, Microsoft MVP, and CNCF Ambassador alumni.
Today, he's obsessed with how AI agents actually write code. At Tessl, an AI agent enablement platform, Baruch focuses on context engineering, management, and sharing. On top of sharing context with AI agents, Baruch also shares knowledge with developers through blog posts, meetups, and conferences like DevNexus, QCon, Kubecon, and Devoxx, mostly about why vibecoding doesn't scale.
Today, he's obsessed with how AI agents actually write code. At Tessl, an AI agent enablement platform, Baruch focuses on context engineering, management, and sharing. On top of sharing context with AI agents, Baruch also shares knowledge with developers through blog posts, meetups, and conferences like DevNexus, QCon, Kubecon, and Devoxx, mostly about why vibecoding doesn't scale.