How Elite Software Engineers Are Using Agents to Get Sh*t Done

Most engineers are using AI coding tools without understanding what they actually are and it's costing them. Microsoft Certified Trainer Rob Bos has trained thousands of engineers on AI tooling, and he sees the same gaps in fundamentals show up again and again, regardless of seniority. This is what you need to know:

• What an LLM actually is (and why understanding this changes how you use it)
• Why prompt engineering isn't optional
• How AI magnifies your existing technical debt instead of fixing it
• The 6-month learning curve nobody warns you about
• Why your role as an engineer was never about writing code
• The environmental cost behind every prompt

Whether you're skeptical of AI tools or already living in agent mode, these are the fundamentals that separate engineers who get real value from those who get burned by the hype.

Connect with Rob:
https://www.linkedin.com/in/bosrob

References:Token tracker: https://marketplace.visualstudio.com/items?itemName=RobBos.copilot-token-tracker
Dev survey: https://www.activestate.com/wp-content/uploads/2019/05/ActiveState-Developer-Survey-2019-Open-Source-Runtime-Pains.pdf

Timestamps:

00:00:00 - Intro

00:00:43 - The #1 Thing Engineers Get Wrong About AI

00:02:09 - How Much LLM Theory Do You Actually Need?

00:03:58 - Why Pair Programming Is Still the Best Way to Learn AI

00:05:26 - Why Rob Skips Tab Completion and Lives in Agent Mode

00:07:03 - The "AI Doesn't Increase Productivity" Debate

00:08:29 - Why Your Real Job Was Never Writing Code

00:09:14 - The 2-Hours-of-Coding Problem No One Talks About

00:11:02 - More Code = More Pressure on Your Review Process

00:12:21 - Why AI Magnifies Existing Technical Debt

00:13:39 - The Customer Who Couldn't Start AI With Developers Yet

00:15:11 - The Future Engineer: Reviewer, Not Writer

00:17:00 - Convincing the AI Skeptic Who Tried It Years Ago

00:19:17 - LLMs Explained Without Visuals (Attention & Semantics)

00:22:41 - Why Prompt Engineering Actually Matters

00:24:20 - From Zero to Hero: The 6-Month Learning Curve

00:26:18 - Is This Confrontational for 20-Year Veterans?

00:29:30 - Becoming a Better Engineer by Thinking in Systems

00:31:26 - Will AI Stop Working as Innovation Slows?

00:34:26 - The Lost Art of Pair Programming with AI

00:35:44 - Tribalism in AI Tools (And Why It's Pointless)

00:37:33 - Tool Agnostic: Start With the Foundations

00:39:40 - Is the IDE Still Relevant?

00:40:50 - The Bluescreen Story That Changed His Mind

00:41:47 - The Hidden Environmental Cost of AI Coding

00:44:15 - 36 Million Tokens in 30 Days: What Does It Mean?

00:45:47 - Running LLMs at the Edge to Cut the Footprint

00:46:48 - Why You Should Be Allowed to Wait Five Minutes Longer

00:47:05 - Outro

#githubcopilot #aicoding #softwareengineering