Lost in the Prompt Sauce? Get Better Results from AI with Context Engineering

Full video transcript

Watch this video to learn how to get better outputs from your prompts without becoming a prompt engineer.

So, unless you’ve been living under a rock for the past four years, you’ve probably heard the word prompt engineering come up in conversation. Someone’s probably told you online that you can just use prompts to make your business or life better. But the majority of people, they’re not getting the best results from their prompts. And here’s the problem. They’re not providing the right context.

Now, you’ve maybe heard, or maybe you haven’t, context engineering. Because ever since prompt engineering was a thing, there was actually a precursor before prompts were even capable of getting you good outputs that don’t just sound like AI garbage, prompt garbage. Look, I’m one to love a good, saucy, well-developed, well-crafted, well-engineered prompt as much as the next person. And in fact, I’ve created a lot of extremely useful prompts over the years that have helped a lot of people do a lot of things with AI. I’m on a level with you.

All context engineering means is giving AI the stuff that it needs to know about you and your business to give you better outputs and results back from it versus a bunch of garbage information going in that gives it no real clear context or outcome. But there’s good news. All you need to do is compile a series of documents that are specific to your company, your product, your service, your business, and who you serve and how you serve them, what makes you different and unique, and having that in a document in one hand and then AI in another — a context engineering sandwich. Super tasty, because prompts with no additional context in generative AI is blah context in, blah context out.

That’s what the savvy folks do, and why I wrote the book called Buyer’s Brief. It helps you make your own version of what this book helps you do, and compiles all of that knowledge and context and information, and helps you consolidate it into a series of documents that can actually be useful for feeding and giving context to AI before you give a single prompt. And of course, there are prompts that can help you generate and compile and organize that context.

But here’s the thing. If you’ve had any sort of training from YouTube University about how to use ChatGPT, and you understand that if you put a really simple, basic prompt in, you’re going to get a really basic, simple output that’s going to probably sound generic, use too many words, use too many em dashes, and probably a ton of emojis that you didn’t want there in the first place.

Now, before we talk about a prompt, I’m going to leave an example of what this prompt will look like to probably the average non-AI-savvy user who doesn’t really understand what ChatGPT is in the first place. And that’s totally cool too. There’s a lot of learning that humanity is going through with understanding how this tech works. Look, the point of AI is to help you with accelerating the results that you are looking to get to faster. But if you don’t even know where you’re going to be going in the first place from the content that you’re going to be creating or using these tools for, then you’re going to have a hard time getting to the right location fast. You might get somewhere fast, but it might be a million miles in that direction.

Okay, so listen. Prompt number one, AI amateur: “ChatGPT, give me a blog about top 10 vets in Austin, Texas.” Okay, ChatGPT: “All right, let’s delve into the top 10, blah blah blah, vets in Austin, Texas.” Okay, you get a basic ask, a basic prompt in, and you get a basic blog post output that probably sounds like generative AI, because guess what — it was written by generative AI.

But if you want to take the fast track to getting a better first draft that actually has more context and relevance to what you’re trying to do, it’d be better to come into the conversation with a simple file or a document or a copy and paste: “Hi ChatGPT, here is all of the information and context and stuff that you want to be aware of about my business. Here is my brand style guide. Here is the tone of voice that I want to be writing in and keep consistent across the board. I want to know what my promise is. What’s the problem that we solve as a business? What are the products and services that we have that actually solve a problem? And who are the audiences that we are serving with this problem? Is it one? Is it two? Is it three? Is it many more than that?”

Well, you want to provide that information to ChatGPT so it understands who you’re writing to, and then also what’s unique about your business, your unique selling proposition. What unique value are you bringing into the world with your product or service? You want to share that information with AI. And look, it can be a framework that you develop, some IP that you have. It can be a lot of different things that you have used in the past to get results for your clients.

And if you’re worried about giving this IP or anything like that to AI — one, if you’re using ChatGPT, you can make sure it’s not training its models based on your information. Go to the settings; it’s very easy to find how to make sure that it is not training those models. But if you’re really concerned about it, then you should seriously consider utilizing tools and platforms that specifically state that they are not using your information to train any sort of models, and that they have the right security and training protocols in place so that you can actually trust that the information is safe and secure. And if you really, really, really want to make sure that this stuff is safe and secure from your business or your ideas, you can create your own local-based models on your computer. Not going to get into that here, but there are ways that you can level up your own security for your information if you want to be super tight-knit about it. There’s also ways that you can just go right into ChatGPT and do it.

And for most people getting started trying to figure out how to get good outputs in the first place, you need to know that the context is going to be the place that you want to spend your time first before you do any of the stuff that I just mentioned.

So back to the example: “Make me a blog. 10 vets, Austin, Texas.” Cool, little context, but still super basic, right? And you’re going to get a basic output from it. But now if you bring in all that information, and then if you provide additional insights and research and things like this, and then maybe you’re a new vet in town, or maybe you’re selling like dog food from these vet places and that’s what your angle is going to be — let’s do that one. So I’m going to write this piece about all the top vets in here, and then I’m going to say, “Hey, if you’re taking your pup to a vet and you want some super healthy dog food, here’s information about our product or store and why should you work with us,” and maybe you’ve already got a relationship with these people and these vets to put your dog food in their place. It’s beside the point, right here. But what I’m trying to get across is that whatever you do, whatever products you make, whatever content you’re looking to create, whatever makes you unique and different, and whatever you actually want AI to know — it might seem obvious to some, but it’s not to most — is that you have to spend the time bringing the context in.

It might take you minutes, it might take you hours to compile all this content. You might be running a business with tons of information across multiple departments. This might be a week-long initiative to have your team go in and actually get all those contexts compiled. But once you do, you keep these knowledge bases as your source of truth that can continue to adapt and be updated, so that whenever you’re going in and leveraging large language models to help you with taking that content and creating content from it that’s contextually aligned, or you’re looking to brainstorm marketing ideas, strategy, do an audit on everything right here — how in the world would you be able to consider AI as a viable resource for auditing your business, your products, your service, your market, if it doesn’t actually know anything about your business?

Let me repeat that one more time, because I see people do this — they literally come into GPT and they give it no contextual information about what they’re doing, what they’re trying to achieve, what problem they solve, who they’re serving, and it gives them some generic answer, because it’s giving the same answer to thousands of other people that are asking the same super simple, not contextually aligned questions for what they’re actually trying to accomplish.

So if you don’t take anything away from this video besides feeling like you might still be lost in the prompt sauce because some guru sold you 10,000 prompts that are going to change your life and business for a dollar online — probably not, let’s be real — but if you are already running a business and you already serve a market and you already have products and services that you can sell to them, whether they’re yours or whether they’re somebody else’s and you’re making a connection as an affiliate or referral, there’s so many ways that you can use these AI tools to help you with doing all sorts of stuff. But if you don’t have the context aligned right out of the gate, everything else you’re going to try and create with these tools is going to be slowed down, and it’s not going to hit the mark the same way that it would be if you just spent a little more time, a little more time going and finding that information, organizing it in a nice series of documents. Throw it in a Google Drive. ChatGPT, Claude, Gemini — they’re all integrating with Google Drive, among a number of other tools and platforms.

Anyways, if all you do is put this context into a Google Drive folder, you are so far ahead of so many people that are like running around with AI tools like a chicken with their head cut off, trying to figure out how to get something out of this when they don’t even have a strategic brief that has all the information that’s pertinent to their business in the first place. And that to me is just a no-no.

Don’t get lost on the prompt sauce before you get your context locked in. Lock in your context, lock in better outputs, better results from your AI content. Just don’t get lost in the prompt sauce before you get that context, right? Context is king. The prompts will help guide the AI after the fact, but if you don’t, you can end up a thousand miles in a different direction than what you wanted to be, with wasted hours, a bunch of AI, and even other technical debt if you’re trying to vibe code some stuff without context. But that is a rant for another video.

I hope you enjoyed this one. Found it useful? Let us know with a like in the comments. If you did, feel free to share with a friend who might find it useful too. And until next time, Darby Rollins from GenAIUniversity.com. Keep calm and AI on. I’ll see you on the other side. Cheers, y’all.