Rajiv Shah – AI Problem Framing for Agentic AI

Here’s What You Get:

Rajiv Shah – AI Problem Framing for Agentic AI (Overview)

Instructor

• Rajiv Shah – AI Problem Framing for Agentic AI
• AI Engineer (OpenHands), professor, speaker
• 10+ years experience, 100+ real-world AI projects
• Known for teaching practical AI thinking, not just tools

Format

• Duration: ~4 weeks
• Time commitment: ~3–4 hours/week
• Format:
  • Live cohort-based (2 sessions/week)
  • Office hours + projects
  • Lifetime access to recordings

Core Idea

Most AI projects fail not because of bad models — but because of bad problem framing.

This course teaches:

• What to build
• Whether you should build it at all
• When to pivot or stop

Instead of coding, it focuses on decision-making and thinking systems.

Key Framework: “The Loop”

The centerpiece is a 5-step framework:

1. Outcome – What success actually looks like
2. Assumptions – What must be true
3. Alternatives – Other ways to solve it
4. Trade-offs – Costs vs benefits
5. Signals – How you know it’s working

This acts like “system design for AI thinking”

What You Learn

1. Problem Framing Mastery

• Ask the right questions before building
• Avoid solving the wrong problem
• Define success clearly upfront

2. AI Diagnostics (Super Valuable)

• Identify what’s actually broken:
  • Data?
  • Model?
  • Architecture?
  • Or framing?
• Use diagnostic tests & signals to decide next steps

3. Decision-Making for AI Projects

• When to:
  • Continue
  • Pivot
  • Kill a project
• Translate AI trade-offs into business terms

4. Agentic AI Thinking

• Understand when agents are the right solution
• Learn the automation spectrum (rules → full agents)
• Avoid overusing LLMs/agents

5. Real-World Case Studies

• 200+ examples of AI successes & failures
• Learn patterns behind:
  • Failed chatbots
  • Misused RAG systems
  • Overengineered agents

Tools & Assets Included

• AI Framing Worksheet (practical tool)
• 5+ strategy canvases & checklists
• Case study library
• Weekly applied project work

Who It’s For

Best suited for:

• AI engineers & builders
• Product managers in AI
• Founders building AI startups
• Tech leads working with LLMs/agents

Especially useful if:

• Your AI works in demos but fails in production
• You’ve wasted time building the wrong thing
• You’re leading AI but lack strategic clarity

Pros

• Focuses on high-leverage skill (thinking), not tools
• Real-world failure-based learning
• Immediately applicable to current AI projects
• Strong for agentic AI decision-making

Cons

• No coding / hands-on building
• Expensive vs typical courses
• Abstract if you don’t have a real project

Bottom Line

This is not a “how to build AI agents” course.

It’s a:

“How to think like an AI architect before you build anything” course

If most AI courses teach execution, this teaches judgment — which is rarer and often more valuable.