What the EdTech?
What the EdTech?
Stop Re-Introducing Yourself to AI
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Most people open ChatGPT, Claude, or Gemini and start from scratch every single time. In this episode, Rob breaks down why that's a trap — and how custom instructions, memory, and a simple AI hack can transform your results in minutes. The bridge between learning AI and living with AI.
Here's something that's been bothering me. I watch people open Chat GPT, Claude, Gemini, whatever whatever tool they're using. And every single time they start from scratch, they tell
The Problem With Starting From Scratch
SPEAKER_00the AI who they are. They explain what they need. They describe the format, the tone, the audience every single time. It's like showing up to work every morning and your coworker has no idea who you are. You shake hands, reintroduce yourself, explain your job title, and then finally get to work. That's what most people are doing with AI right now. And it's completely unnecessary. Let's start with something everyone's heard by now. You need to learn prompt engineering. And look, that's that's not wrong. When you first start using AI, the way you ask the
Why Prompt Engineering Isn’t Enough
SPEAKER_00question is everything. The difference between typing write a lesson plan and giving detailed context about your grade level, your standards, your format preferences, it's night and day. But here's the trap people hear prompt engineering and think that's the permanent skill. That forever the game is writing better and longer and more detailed prompts. It's not. Prompt engineering is the starting line, it's how you get through the first few weeks. But if you're still writing five-sentence prompts six months in just to get a decent response, something went wrong. You missed a step. And that step is custom instructions. So what are custom instructions? Simple. They're a set of persistent directions you give to the AI once, and they
Custom Instructions Explained
SPEAKER_00apply to every conversation going forward. Think of it as onboarding a new team member. You sit down, you explain, here's who I am, here's what I do, here's how I communicate, here's what I never want to see. And after that orientation, they just know that's what custom instructions do for AI. Without them, every chat starts at zero. The AI doesn't know you're a superintendent, doesn't know you prefer bullet points over paragraphs, doesn't know your district uses the 5e model, nothing. So it gives you generic, watered down, one size fits all output. With them, your first message can be six words, and the response is already tailored to your role, your audience, and your standards. Here's where it gets interesting. Every major platform has a version of this now. Chat GPT calls them custom instructions, found under personalization in your settings. Claude uses projects, you drop in instructions, reference
Platform Options And Names
SPEAKER_00documents, even style guides. Gemini uses gems, same concept, packaged a bit differently, and perplexity has spaces, where you define the AI's behavior for specific workflows, different names, same power move. Now here's the part that really matters, and it connects back to something I've been talking about in previous episodes. Custom instructions don't just save you time in one conversation, they create a
The Compound Effect Of Setup
SPEAKER_00compound effect across every conversation you have with that tool. Think about what happens. You set your instructions. Now every response is slightly more aligned with how you think, how you write, how you work. You refine those instructions over time. Maybe you realize you want the AI to always suggest counterarguments, or never use jargon, or always include citations. Each refinement makes every future interaction better, and then layer in memory. The major platforms now remember things about you across sessions. ChatGPT builds a rolling profile based on your conversations. Claude just launched persistent memory for paid users, and it shows you exactly what it's stored, so now you've got custom instructions setting the foundation and memory filling in the gaps. Day one, you're writing a detailed paragraph
Step-By-Step: Generate Your Instructions
SPEAKER_00just to get a mediocre result. Month six, you type one sentence and get exactly what you need, not because AI got smarter, because it got to know you. Alright, here's the practical part, the how, and this is the hack that most people completely miss. You don't have to write your custom instructions from scratch. You use the AI to write them for you. Here's exactly how to do it. Open your preferred AI tool and say something like this. I'm going to give you my LinkedIn profile, a few examples of my writing, and a description of my role. Based on this, generate a complete set of custom instructions I can use so that every conversation is tailored to my work, my audience, and my communication style.
Translate Across Platforms
SPEAKER_00Then paste in your LinkedIn bio or about section, two or three emails or documents that represent your voice, and a short paragraph about what you typically use AI for, drafting communications, analyzing data, building lesson plans, whatever it is. The AI will generate a full instruction set, your role, your tone, your formatting preferences, your audience, topics you care about, things to avoid, all of it. From there, take that output, open your settings in whatever platform you use, and paste it in five minutes, maybe ten if you want to refine it. And here's a bonus move. Once you've done it for one platform, ask the AI to adapt those instructions for each of the other platforms. ChatGPT's format is slightly different from Claude's projects, which is different from Gemini's Gems, but the AI can translate between them instantly. Let me zoom out for a second. Forbes published a piece earlier this year that said prompt engineering is no longer the most valuable AI skill in 2026.
The Real Skill: Infrastructure Over Prompts
SPEAKER_00And I think that's right, but only half the story. Prompt engineering isn't dead, it's just evolving. The valuable skill isn't writing a perfect prompt every time. It's setting up your AI environment once so that you rarely need a perfect prompt. Custom instructions, memory, projects, spaces. That's the infrastructure. The people who seem like AI wizards, the ones getting incredible results from one-sentence inputs, they didn't skip the learning curve. They automated it. They front-loaded the work into their setup so that every conversation after that starts at level 10 instead of level 1, and that's available to everyone, right now, for free on most platforms. So here's my challenge to you. If you haven't set up custom instructions yet, do it today, not tomorrow. Today, open your AI tool, give it your bio and your writing samples,
Your Two-Week Challenge And Payoff
SPEAKER_00and let it generate your instructions, paste them in, start using the tool, and watch what happens over the next two weeks. Your prompts will get shorter, your outputs will get better, the friction will disappear. Custom instructions are the bridge between learning AI and living with AI. They're the difference between a tool you use and a tool that works for you. Set them up once, refine them over time, and let the AI do the heavy lifting of remembering who you are so you can focus on the work that actually matters. That's the move. I'll catch you in the next one.
