Read time: 3 minutes Hey Reader, It's been a while since I last wrote this newsletter to you. The last few weeks have been hectic, and I only now have some time to sit down and focus on teaching. 4 steps to succeed in AI in 2024With over 12,000+ AI tools online, and all these new language models that are coming out, it might feel hard to stay updated. Let alone if we think about the new image and video generative models, voice cloning tools and use-case specific tools like AI integrations in pretty much every software you already use (Slack, Notion, LinkedIn, Instagram, Zapier, etc. just to name a few). So in this email, here is the 4 step shortcut to get caught up with AI ASAP. 1/ Learn Prompt EngineeringText generative models' performance (like ChatGPT, Claude, Gemini, LLaMa, etc) are determined by 3 key factors.
If we are only talking about using LLMs in the browser interface and not integrated into systems, you can't control 2 out of these 3 variables. Because the model's architecture and training data is given and there's nothing you can do about it. Extensions are also a given, controlled by the publisher of the model. So the only thing you can really control is how you give commands to these AI tools, and how well you can decide what tasks to use them for. Just like how a colleague can't read your mind on what you want; you have to effectively communicate with other people when you are asking for help; an AI can't read your mind either (not yet). Prompt Engineering is a skill of effective communication with AI Language Models. It's the art and science of writing commands that can get you the outputs that you want, reliably (more on this later). If you want to learn it and master it, enroll in the Prompt Master AI Course today! 2/ Master Conversation DesignOnce you get the hang of Prompt Engineering, you will not only see its strengths, but also its limitations. For example, that you cannot write a prompt that writes you a book, no matter how hard you try. Simply because of the limitation of the model's outputs. They can't output that much content once. Another example problem you can't solve with prompt engineering, is if you need to make a 10-20 page proposal of a project where you have to include many different sections, that would require totally different prompts to ensure high quality. These problems are where we have to use a Conversation Design technique where we break down the problem into smaller parts, and write our prompts one at a time, getting the AI model to not only focus on one task at a time, but also feeding its context window, so the next prompt can build on top of the previous responses. Conversation Design is how you can get outputs out of AI Language Models otherwise impossible with a single prompt. If you'd like to learn about the 6 beginner and 8 advanced Conversation Design methods that I teach and use every day, enroll in the Prompt Master AI Course. 3/ Configure a Custom GPTOkay, once you know how to get any kind of input from ChatGPT, whether long or short, it's time to write prompts, and configure custom versions of ChatGPT (called Custom GPTs), that are highly specific for one single task. Here are a couple of use cases that my students and I put together:
And the list goes on and on. How well and reliably these GPTs perform is absolutely based on how well you write the Instruction prompt, because in a Custom GPT, you only have 1 shot to write a good prompt. And once you can configure a Custom GPT that works well, safely and reliably, you can also configure a chatbot to run on your (or your client's) website. It's pretty much the same thing. One well written prompt, and maybe some extra knowledge base connected to it. If you want to learn how to create chatbots that are user-friendly and work safely and reliably, enroll in the Prompt Master AI Course, where I'll walk you through it, and you can also watch me critique other students' GPTs, and give feedback on how they could improve them. 4/ Build no-code AI AgentsOkay so let's say you are amazing at configuring these custom GPTs and you can make an AI do most of your annoying tasks in your business or job. Then you face the next challenge: I still have to open ChatGPT and prompt it, or trigger my CustomGPT somehow. So a custom GPT is little bit like an employee you have to micro manage, and tell them what to do every single time. Enter the next, and final level: building AI agents. These agents can:
Okay, I know that sounds super vague. Let me give you an example: Handling Customer Support Tickets
Again, HOW WELL an AI agent works all comes down to how well you can put your thinking and human-based customer support process into one or multiple prompts that handle this smart decision making. As a Make Academic Alliance Partner, we use Make to build no-code AI agents for ourselves and our clients. If you want to learn exactly how we do this, enroll in the Prompt Master AI Course, and you'll also get a bonus 3-month free membership on the Make Core plan, only available to our students. A lot of our students found success in this course, and this is a course I also taught at 2 universities this year over a 12-week curriculum. Once you join, you'll also get lifetime access to the Promptmaster Community, where you can meet the other students, ask questions, join our live Q&A calls, get support for your projects and ask questions directly from me. The entire course is designed to walk you through these 4 steps that I outlined in this email, so if this resonated, you should totally check out the Prompt Master AI Course, and join! But regardless of how you decide, I'll stay in touch with more educational content, just like before. Best, Dave Talas P.S: In the Prompt Master AI Course, you can master the 4 steps you need to get caught up and ahead of most of your peers and competitors. Enroll today! |
Sign up if you are overwhelmed with all these AI tools out there, and want a clear path on your AI journey.
Read time: 3 minutes Hey Reader, It is one thing to know the potential of AI Language models like ChatGPT or Claude, but it's also important to know their limitations. In this email, I want to show you 5 things you should avoid when using ChatGPT (or Claude or Gemini or LLama, etc.) 1/ Treat it like a know-it-all search engine ChatGPT is not a search engine. It is simply a very expensive predictive keyboard. All it does is it just guesses what's the best next word to say in a scenario. It...
Hey Reader, One of the most common requests I hear from clients is that they want a custom version of ChatGPT for their business that uses their knowledge. Whether it's a chatbot on their website, WhatsApp, Instagram, etc. all they want is to have it use their knowledge. So in this email, I want to teach you how to build such a thing. I just created a custom chatbot for my website in 40 minutes, you can give it a try on this website (chatbot bubble in bottom right corner). It has access to...
Hey Reader, Today, I wanna showcase a huge win from a student, who saved over 100 hours with an AI automation he built. He shared this with our community last week and I think it's one of the biggest student wins that we know of. So here are Tudor's words (my highlights): The challenge: Introducing 450 new products to an online store with only the product photos available. No descriptions, no meta tags, no attributes—just images. Our solution: A custom automation tool that analyzed each image...