Hey, my name is Mike and you're listening to Lone Wolf Unleashed, the podcast where I show you how to live larger and switch off sooner.

This week we're going to be talking about being AI prepared.

There's a lot of conversations I've been having recently, things about making the implementation of AI simple.

A lot of the conversation comes back to how well prepared people are for the implementation of AI.

The answer to that is both it is simple and it's complicated.

It's simple in that the concepts are very, very simple to understand.

And I'm gonna walk you through that today.

It's complicated in that there's actually a lot of work that needs to happen before you're really ready.

So I'm gonna walk you through now what the concepts are that you need to wrap your head around so that you can get ready to use AI to its fullest potential in your business.

And then we're going to have some actionable steps at the end that is going to help you implement well so that you can start to save time.

So if you listen to this, you've probably already experimented with artificial intelligence, large language models, generative AI, whatever you want to call it.

You've probably experimented with things like ChatGPT or Claude or Gemini.

There's a million AI tools out there now.

And it does feel like that we're in a little bit of a AI tech bubble.

It's a bit like the Internet boom of the 90s.

While we're here, how is it that we can start to leverage this new technology?

So if you've had a play with this technology, you'll know that it can help you brainstorm things and ideate things and help you sort of write faster or think faster or whatever it is.

The premise of it is that it should be able to save you time in doing tasks that were relatively mundane before.

There's a few attitudes that we need to sort of tackle before we get into the tactics.

The primary attitude here is in deciding what task AI is going to take over, what is the purpose of that task?

And the lens that I look through it with is, is this a human centric task or is this a machine centric task?

So previously, the whole idea around automation has been let the machines do the machine work and let the people do the people work.

And that is still true of AI.

And there's a few mistakes that I feel people are making here in the implementation of certain AI systems, particularly around ones that do phone calls and voice generated stuff.

I feel like people are better at doing relational things than machines are.

I might be crazy in saying that the AI bros are going to come after me for saying that.

What I'm trying to help people think through is before you even get to voice, there are so many things you can do in your business that will give you process and productivity efficiencies before you get there with the lens and the attitude around the types of tasks that things can take over for you.

Don't let it take over your strategy.

Strategy is complex, it is difficult, it requires a lot of nuance.

Don't let it interact with your customers without having the proper security and controls in place.

Okay.

I find it really difficult where people will talk about getting AI to do end to end automation when they don't even understand how the process is supposed to work in the first place.

For example, there's some companies recently I've heard about using AI to make calls.

How they do it is they only will get the AI to take over follow up after the human agent has not been able to get through three times.

So there's a follow up sequence and the AI agent's primary goal is to set an appointment.

That's it.

It's not trying to sell, it's not trying to do anything other than secure a time for the prospect to talk with a salesperson.

Now that is a far more appropriate use because you're putting humans first, but you're also not trying to get the AI to make the sale because what goes into making a sale is really complicated and there are many thousands of different buyer journeys that a buyer can go through.

The complexities around that particular buyer situation.

And you know, if you work through the Bant framework, budget, authority, need time, that's going to be different for every client.

So having a dedicated salesperson is actually really valuable there because they can understand and they can operate in ambiguity.

Problem with an AI system is that it finds it very difficult to operate in ambiguity.

And it is a piece of technology, it is a piece of software, it deals with black and white rules and, and so the amount of context that you need to give it to operate in the same capacity as a person would be massive and it would chew through so many tokens it would be very, very expensive to run.

So before you sort of decide what sort of task do you want to give AI, just think about what impact will implementing this have on my business and what risk does it pose to my business if it gets it wrong.

I'm trying to caution people not to get caught up in the hyp now we can turn to how do we get ready?

So there's two foundations that we need to lay here based on the type of tasks that we want an AI agent to take over.

They both end in bases.

So there's the two bases, okay?

There's the first one, which is the knowledge base, and there's the second one, which is the database.

So there's two types of work that AI initially can take care of for you.

The first one is knowledge.

If you have a knowledge base, you can query the knowledge base through the AI agent and it can provide you a response.

It can give you, in time or just in time, information for you to be able to complete a task that you're doing.

It can sort of direct you to do the right thing that's next along a particular process.

And that's very useful.

The work that goes into that is gonna be different to what you do for a database.

The second one is the database.

Now, database is the system behind that's connected to the AI agent for the AI agent to be able to communicate with and to do things in.

So I'll give you an example.

I use an ASANA workspace, and in there is a series of projects.

What I sometimes do is I'll be talking with Claude and I'll be saying, hey, this thing that we've been talking about, can you now create for me a series of tasks in Asana so I can go and keep track of completing those things?

So what it will then do is, because I've got ASANA connected to it, it will go and it will add things to a database, to a task database in Asana, for me to keep track of.

It doesn't need a terrible lot of knowledge to be able to go and do that.

It will literally just create the task like it's been told to.

Now, how do we bring all of that together?

So how do we combine a knowledge base with a database with an AI agent to supercharge the outputs that it can do and the quality of those outputs?

So if I have a process around how I manage a project, there's going to be understanding the requirements of that project, there's going to be setting up the project, all those things.

If I have documented the knowledge into a knowledge base and then I've connected that with context around how it should act with a database, it is now going to give me a much better result.

So it might change the way that it names tasks.

Okay, I haven't prescribed that before, but the way that you name a task is not a Database activity.

The way that you name a task is through a knowledge base.

It is a knowledge activity.

So it's something I could show someone else how to do.

It's knowledge that they would then have, that they can then go and act on.

Knowledge is about knowing the database is about the doing, it's the operational side, it's the execution of the knowledge.

So you might be sitting here and go, I don't even have my knowledge documented, I'm starting to use AI.

And you might go, well, I already know how to do my job, I'm not going to bother documenting that out.

But I'll tell you this, it is more pertinent now, it is more important now more than ever to document the knowledge, even if you are a solo operator.

And that is because we are now able to automate just based on giving English based prompts to piece of technology.

Something that's never been able to be done before, we can do now.

Part of that English sort of plain language input is the knowledge on how to do that task and how it's supposed to be done and how the output is supposed to look.

So that includes doing up your templates, doing up how to use those templates, doing up how you know to name them and what fields need to be put where and all of those sorts of things.

If you provide all of that context to an AI agent, you should be able to get a high fidelity result out of that.

I'm going to tell you now about how I'm doing this with my own process.

Part of what I do with clients is I model out processes and we understand what activities need to happen in what order and who does them and what the dependencies of them all are and all those sorts of things.

And it really helps us understand what improvements we can make in the future.

Three years ago I would sit in a room with someone and I would charge by the hour.

I would model out a process while people talk to me.

Especially when you're doing this with a group of people, if you're doing it with a team, it can be quite a tough gig to facilitate and model and pay attention and make sure I'm asking the right questions and all that sort of stuff all in the one session.

Now I have the ability to take a transcript, I have a custom prompt that I have with Claude where I've told it how to treat different process language, I've told it what formats to use, I've taught it how to model in BPMN properly because the background of BPMN is Just an XML file.

So it's just a piece of code.

What that means is that given all that context and that knowledge, and I've taught it how I model and I've taught it how my attitude around different levels of detail, I can now take a transcript from a call.

The output of that is a high fidelity process model based on the conversation of that call.

It means that for the most part I don't have to model the initial process anymore, which means I can now spend a higher amount of my time in those conversations asking really good questions and paying more attention to what the person is saying rather than the thing that I'm also doing on the side.

Overall, what it allows me to do is to do a much better job with a better outcome for the client because I'm able to start to recognize some of these opportunities for improvement where I may have previously missed them because I had my focus being split across two things.

Now that's a fairly sophisticated example.

I'm not saying that the first thing that you're going to go and build into an AI agent is something that writes code for you and displays it in a process model type.

That's not the point.

The point is, is that bringing together how knowledge works and giving it a context so that the agent can act for you, that is the difference maker in terms of getting AI to make things faster for you, rather than just being a hype tool that people are building up to be the second coming of Christ.

What you want to be able to do is you want to be able to feed it your knowledge so that you can get a really good outcome.

The cool part about this though is this.

The old methods of documenting knowledge, so procedures, checklists, templates, they will all.

So even if you document those things, and let's say you had the capacity to do that tomorrow, then the use of those things alone will increase your productivity without an AI agent.

I've just seen this with a customer of mine.

In developing out the process, in developing out their related artifacts to make them faster to get agreements on what is a case and what's not a case, when the process should actually run versus when it shouldn't, we've seen in some parts of their process a five times improvement in efficiency, a 500% increase in efficiency, or to put it a different way, a 500% increase in efficiency.

That is amazing.

And that's without an AI.

If we were to take their templates and how to fill in those templates with the knowledge docs and their checklists, we would Be able to have an AI agent be able to create the letters, the contracts, the whatever it is within a very short period of time.

It makes implementation exponentially easier.

Even by not putting in an AI agent, you're getting those benefits.

Now you can supercharge those benefits by then building it in because it has those contexts.

So have a think about how you can do that for your business.

Have a think about if I'm going to give an AI agent something to do, whether that just be a simple project that you set up in Claude or ChatGPT, or whether that be a custom workflow that you're interacting through, N8N or one of these other tools, one of these other automation tools.

What knowledge would I need to give it for it to give me the best outcome?

So treat it like a person in that respect.

Treat it like someone who has never done the job before.

What would they need to know if you were going to give them a task on how to do the best job possible for you, I highly recommend you do that.

If you do that, you will be ready to implement AI agents really well.

You can stop talking with ChatGPT and you can actually start doing real AI agentic automation, which is super exciting because if leveraged correctly, business owners out there who are overloaded, who have made the sales but really need to deliver, and they walk the line between getting more sales and delivering quality.

They don't need to decide or have a choice on that anymore, they can do both.

That's super exciting for me and I think that should be really exciting for you as well.

Because I see you, I see you working those 60, 70 hour weeks and I think that even though you might get fulfillment out of what you do, there is more to life than work.

My desire is that you are prepared, you are equipped to implement things in your business, implement systems in your business that work, that have a higher chance of delivering on the benefits that they promise.

So that is the episode for this week.

Thank you so much for listening today.

You could have been doing a million other things but you decided to come and hang out with me and learn how you can be ready to start to implement AI agents in your business.

And for that I want to thank you for you and your time.

Thank you so much, so much.

Thanks so much for listening.

I'll see you in a fortnight.