Visual summary of operating lessons from Sander Schulhoff.

Lessons from Sander Schulhoff

The items below are things Sander Schulhoff said in three podcast interviews about prompt engineering, AI security, agents and learning AI skills. — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.

Part 1: The Prompt Report & Research

  1. On Bad and Good Prompts: On Lenny’s Podcast, Schulhoff said: "using bad prompts can get you down to 0% on a problem, and good prompts can boost you up to 90%". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  2. On Examples: On Lenny’s Podcast, Schulhoff said: "Examples are a common element in prompts, but giving examples is also a prompting technique". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  3. On the Directive: On Lenny’s Podcast, Schulhoff said: "The directive is a part of a prompt, and that's your core intent". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  4. On Product-Focused Prompts: On Lenny’s Podcast, Schulhoff said: "the most important places to use those techniques is the product-focused prompt engineering". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  5. On Millions of Users: On Lenny’s Podcast, Schulhoff said: "With product-focused, millions of users are interacting with that prompt". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  6. On Cost and Accuracy: On The Cognitive Revolution, Schulhoff said: "I try to put as many as is reasonable in my prompt, weighing costs with accuracy and speed". — Delving into The Prompt Report, with Sander Schulhoff of LearnPrompting.org, The Cognitive Revolution.
  7. On Structured Input and Output: On The Cognitive Revolution, Schulhoff said: "what I always do is have a structured input output, as I said, the Q A, and that's what I've seen to be standard". — Delving into The Prompt Report, with Sander Schulhoff of LearnPrompting.org, The Cognitive Revolution.
  8. On Few-Shot Prompting: On The Cognitive Revolution, Schulhoff said: "Few-shot prompting, giving it examples, super, super important because you can't always describe in words exactly what you mean". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.
  9. On Human-Written Examples: On The Cognitive Revolution, Schulhoff said: "it's not as good as human-written examples for the most part". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.

Part 2: AI Security Fundamentals

  1. On Unsolvable Problems: On Lenny’s Podcast, Schulhoff said: "It is not a solvable problem, which I think is very difficult for a lot of people to hear". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  2. On Never Being Certain: On Lenny’s Podcast, Schulhoff said: "you can never be certain with any strong degree of accuracy that it won't happen again". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  3. On Social Engineering: On The Cognitive Revolution, Schulhoff said: "It might not even be a solvable problem at all, because if you look at humans, this is analogous to social engineering". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.
  4. On Serious Problems: On Lenny’s Podcast, Schulhoff said: "Governments are looking into this strongly, security and intelligence communities, so it's a really, really serious problem". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  5. On Detecting Scammers: On Lenny’s Podcast, Schulhoff said: "The reason that we're so able to detect scammers and other bad things like that is that we have consciousness and we have a sense of self and not self". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  6. On Model Creators: On The Cognitive Revolution, Schulhoff said: "I really think this is a problem that's going to be solved by the model creators themselves". — Delving into The Prompt Report, with Sander Schulhoff of LearnPrompting.org, The Cognitive Revolution.

Part 3: Prompt Injection & Red Teaming

  1. On Improving Your Prompt: On Lenny’s Podcast, Schulhoff said: "The most common technique by far that is used to try to prevent prompt injection is improving your prompt". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  2. On Red Team Incentives: On Lenny’s Podcast, Schulhoff said: "if you look at contracted AI red teams, maybe they get paid by the hour, not super incentivized to do a great job". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  3. On Safety-Tuning: On Lenny’s Podcast, Schulhoff said: "Fine-tuning and safety-tuning are two particularly effective techniques and defenses". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.

Part 4: The Failure of Guardrails

  1. On Guardrail Models: On Lenny’s Podcast, Schulhoff said: "A lot of times the guardrail model won't even be intelligent enough to understand what that means". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.

Part 5: Prompt Engineering Best Practices

  1. On Learning by Trying: On Lenny’s Podcast, Schulhoff said: "You will learn the most from just trying and interacting with chatbots, and talking to them, than anything else, including reading resources, taking courses, all of that". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  2. On Boosting Performance: On Lenny’s Podcast, Schulhoff said: "just by giving examples of what you want, you can really, really boost its performance". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  3. On Structuring Prompts: On The Cognitive Revolution, Schulhoff said: "They knew how to structure them properly and were able to get a lot more benefit out of using AI". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.
  4. On Leaderboards: On The Cognitive Revolution, Schulhoff said: "as far as leaderboards go, they don't always tell the full picture". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.

Part 6: Agentic AI & Advanced Capabilities

  1. On Frustrating Agents: On The Cognitive Revolution, Schulhoff said: "Agents right now are quite frustrating because it looks so promising, but in practice, it's so hard to get stuff working even for me". — Delving into The Prompt Report, with Sander Schulhoff of LearnPrompting.org, The Cognitive Revolution.
  2. On Agent Designers: On The Cognitive Revolution, Schulhoff said: "I think a lot of companies will be hiring agent designers". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.
  3. On Toy Projects: On The Cognitive Revolution, Schulhoff said: "that fun toy project is really great to understand how they work, and there's a lot of nuance in designing these". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.
  4. On High Salaries: On The Cognitive Revolution, Schulhoff said: "you're not just doing prompt engineering if you're getting paid $300k". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.

Part 8: Education & The Future of AI

  1. On Persistence: On Lenny’s Podcast, Schulhoff said: "persistence is the only thing that matters". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  2. On Loving Math: On Lenny’s Podcast, Schulhoff said: "I'm really not very good at math, but I love math, and love AI research and all the math that comes with it". — AI prompt engineering in 2025: What works and what doesn’t | Sander Schulhoff, Lenny’s Podcast.
  3. On Training Employees: On The Cognitive Revolution, Schulhoff said: "companies are going to train people in their company to learn generative AI skills". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.
  4. On Advancing Capabilities: On The Cognitive Revolution, Schulhoff said: "capabilities are advancing, have advanced, and likely will continue to advance pretty rapidly". — Ignore Previous Instructions & Listen to This Interview | Sander Schulhoff, CEO, LearnPrompting.org, The Cognitive Revolution.