@tjrobertson52

If you're like most businesses, you're already using AI to do some work in your business. You've identified some set of tasks that you can hand over to AI, maybe you've even created some skills. But you also get the sense that AI could be doing so much more.

In this video I want to talk about how you can do that. Our agency is still pretty people heavy, but most of the hands-on work is done by AI. I mean like probably 70 to 80 percent of the hands-on work is done by AI, mostly Claude, and within a year I expect that to be 90 or 95 percent.

And that's not because we're delivering a bunch of AI slop. The rule in our agency is we only give a task over to AI if AI can truly do it better. It's not enough that it can nearly do it faster.

We just devote a lot of time and resources into integrating AI into our processes. And by the way, a year from now, I still think we'll be very people heavy. People tend to multiply the value produced by AI.

So the more value AI is producing, the more valuable humans become. To do this right, you need to go from giving AI discrete tasks to fully integrating AI into your business, or maybe more accurately building your business.

It's the difference between onboarding a junior level member of your team that has to be given step-by-step instructions and onboarding a senior level member of your team that just needs to be given context and goals. Setting this up for AI requires two parts, data and skills.

Data is all the hard facts and information about your business. I recommend storing this information following something like Google's OKF standard and keeping it in a place like GitHub so it's easy to integrate with any AI. I have lots of videos talking about that.

We build this for all of our clients. We call it their brand ambassador. What I really want to spend time on in this video is skills, because the way we create skills for new models is changing.

If you don't know, a skill is just a set of Markdown files with instructions for AI. Except, like I said before, the newest, smartest frontier models aren't like low employees that need step instructions.

6, you shouldn't be telling it exactly how to do the task. You should just be providing all the context it needs to work as a representative of your business. Again, all the context that a senior level member of your team would need to pursue your goals.

So let me give you some examples. Our team uses Notion for task management. So we need one comprehensive skill that spells out how to do anything in Notion.

Where everything is, what every single field means and how all our internal processes work inside of Notion. And then anytime Claude needs to do something inside of Notion, it can refer to that skill. And here's the important bit.

No other skill should ever talk about where things are in Notion. This is an important principle to keep in mind as you start to build up these skills. In coding, it's called DRY.

Don't repeat yourself. Each piece of information should have one home.

If any other skill needs that information, it should just reference that skill. But Claude needs other types of information as well, like our general philosophy around SEO. So we need a skill all about website optimization.

A skill all about authority and how that impacts rankings. A skill about keyword research and how to pick the right topics for content. None of these skills are step-by-step instructions telling Fable exactly how to do something.

Fable is entirely capable of figuring out the best process to achieve a given goal. In fact, if you do spell out step-by-step instructions, you're likely just hamstringing it, preventing it from coming up with a better process to achieve that goal. What it needs is context and data.

Context around how your specific business operates and thinks. It takes a long time to document and organize all that data, but it's well worth it.