🔥 Credentials Are Dead — In a world of AI abundance, it's not what you know on paper, it's what you can do that matters.

🧰 Meta Prompting, Mastered — A powerful framework to prompt your LLM to… write better prompts.

🎯 Certifications, AI Agents & Reality Checks — Why OpenAI's training push matters, and why most AI agent projects are built on shaky data foundations.

Having credentials is going to be irrelevant in the near future. The only differentiator is going to be if you can actually do the work at hand and adapt quickly. We, as a society, are largely to blame for this. Grade inflation, falling standards, and the willingness to accept mediocrity are the reasons for this. Diplomas aren't worth the paper they are printed on.

This isn't a bad thing. Credentials used to be an easy way to determine competence and experience. Now they aren’t. You can’t assume that because someone has a master’s degree or a specific certification that they have the subject matter expertise and the ability to apply it. It’s frankly easy mode to try a certification because you get a syllabus and plan, but that is so far from the real world. You're not given a syllabus when you start a job, and the skills need to succeed in the AI world are vastly different than what is being taught in college.

The risk is that the divide between the haves and the have nots is going to widen, and AI accelerates this trend. The people who are the best at what they do are going to get most of the gains. Each industry is going to resemble Hollywood, where are few of the top actors take home millions, but where the average in the US makes $56,000. The question is what happens when more and more people are sidelined in their job search? AI is helping mid to senior professionals, while augmenting and sometimes replacing entry level work. There may not be a talent pipeline at all if we don't hire entry level workers.

Bill Gates says goodbye to doctors — Bill Gates suggests AI could make doctors “irrelevant” for many cases. Is it visionary... or too soon? Either way, the disruption is real. Read more →

AI plays job market middleman — In a twisted loop, companies are using AI to reject candidates, while job seekers use AI to apply. Humans optional. Read more →

OpenAI: Embrace the 'I don't know' — New guidance tells models to admit uncertainty instead of hallucinating. A big step toward trust and fewer LLM-induced headaches. Read more →

Why take so much time trying to think of and perfect your own prompt? Let the LLM create and refine a prompt that you can then feed back into it. This is called meta prompting. It's a powerful technique, and as you use it more I've found it works really well if you're struggling to refine your prompt and can help you tailor your prompts more effectively. Here is an example below:

"I'm trying to learn more about AI and machine learning. Can you help me design a prompt that I can use that will effectively teach me? I want it to follow this framework: Role -> Tasks -> Context -> Format -> Constraints -> Iteration."

OpenAI is disrupting the job market with building out their Open AI Jobs Platform. They are offering an AI certification, which may be the rare certification that is worth it to get. They've committed to having 10 million Americans certified in AI by 2030. It's a great way to speed up the reskilling/upskilling, and you can bet that it will adapt as new techniques are discovered. There is also the OpenAI Academy, that has a combination of different learning opportunities and events.

And before you think I'm getting paid sponsorship from OpenAI, other major LLM companies are doing similar things. There is a walkthrough on how to build AI agents from Anthropic's site. Google has Gemini LearnLM, which is an AI assistant that can accelerate how you learn. I still hear people tell me today they don't know where to start or it is difficult to learn, but a simple 30 second search found these resources. The information is all out there, and the barrier now is the just the will to step outside of your comfort zone and be willing to feel like a novice.

AI agents are much more challenging to implement and govern in real life than any vendor or influencer will let on. The message is usually more agents = make more $$$$, and all the tough parts about having a data strategy and well governed data isn’t touched on. This is a huge mistake, and there is a going to be a big hangover when companies realize the millions they are spending on AI agents would be better allocated upstream to improving their data quality and governance.

And I get it. It's difficult to get people excited about data quality and governance. If you're lucky, there is neutral sentiment for these topics, and in most cases it skews slightly negative. What you'll find those, with AI agents or AI in general, is that eventually you have to come back to the data foundations. Data will always be the limiting factor, and if you focus on that first, you'll accelerate your AI journey. The question is, would you rather do the hard work now, and have to do it in 3 years and basically pause your AI efforts?

Agentic AI is going to follow an accelerated Gartner hype cycle, and I would argue that there will be some mini cycles built into it. It is a fast moving part of AI and many of the processes and standards are being made on the fly. Like traditional AI before it, Agentic AI shines where tasks are repeatable, rules can be created, and autonomy adds efficiency or scale. There also needs to be actual documented processes, which many companies struggle with. So if you really want to apply AI agents to your internal business problems, focus on the ones that meet this criteria, and you can implement it in a sustainable way.

-Ylan