Good technique: frontloading

I watch a lot of students and business professionals using AI, and while there are some who totally get it right away, for most—it’s like watching someone try to eat soup with a fork. Sure, you'll get a little soup into your mouth, but it really isn't optimized. What’s the big mistake? They just dive in headfirst without a plan.

Now, if you’re treating AI like a fancy search engine, that’s fine—it actually works just fine by diving straight in. But if you’re trying to create something complex, like a large proposal or a complex brand strategy, that approach is like trying to build IKEA furniture without the instructions. You’ll end up with a wobbly mess and a pile of leftover screws.

But this actually makes sense because in at least one way, AI's are a lot like humans—they need lots of context to perform well. Context, for me, is at a minimum answering the “5 W questions” (what, when, who, where, why). If you don't start with basic information that the 5 W's begin to answer, it has no idea what you want, and you’ll end up with results that are vague, generic, or just plain unhelpful.

Broadly speaking, there are two kinds of tasks you can tackle with AI: “context-rich” and “context-poor.” Context-rich tasks are the jackpot. Let’s say you’re working on a response to a well-written RFP. You’ve got the document itself, a link to the client’s website, the LinkedIn profiles of all the execs, maybe even news articles about the company or its industry. This treasure trove of information is exactly what your AI thrives on—as long as you give it the info strategically. That’s what I call frontloading: feeding your AI all the juicy details upfront so it can actually help you create something tailored and brilliant.

AI gets really overwhelmed if you throw too much at it all at once. If you try to dump everything in at once, it’ll “choke”—basically, hit its memory limit and give you unusable fluff. To avoid this, you need to break the information into manageable chunks. I call this technique "chunking" and it works like this: feed the AI bite-sized pieces, confirm it understands the context (without repeating back the results (this is key) and then move to the next chunk. This keeps it focused, reduces the risk of wandering responses, and dramatically improves the quality of its output. Yes, it takes a little more time upfront, but the results are so much better—and it’s still way faster than starting from scratch.

If you want to dive deeper into techniques like frontloading and chunking, let’s chat! I offer one-on-one and group courses that can take your AI skills from “pffftt” to “whoa.” Reach out, and let’s build something amazing together.