161 | AI Projects Best Practices - Step by Step Guide For Planning and Executing Projects with AI and Existing Tools in Your Tech-Stack with Josh Huilar

0:00Hello, everyone, and welcome to another live episode of the Leveraging AI podcast, the podcast that shares practical, ethical ways to leverage AI, to improve efficiency, grow your business, and advance your career. This is Isar Maitis, your host. And we've got an interesting and different topic for you today, but I think it's very, very important. It was always very important, just becoming more important. And businesses are run around projects. Some are small projects. Let's figure out how to throw a birthday party to Michelle from accounting. Some are huge projects like let's set up a data lake house storage for all of the company data and everything in between.

0:40But regardless of the size of the project that you're running, every project requires a deep understanding of the requirements and then workforce planning. In other words, how to allocate specific resources to achieve the required outcome in the most efficient way. Now, that was always true. That's not new. Like I said, every single aspect of a business is some kind of a mini project and some kind of very big projects. But now this topic becomes a lot more critical because of the introduction of AI. And what I mean by that is soon as AI becomes more and more implemented across all aspects of the business, and then we're going to have agents, and then we're going to have AGI, and then we're going to have ASI, the workforce will continuously and dramatically change from what it is right now.

1:26Meaning knowing how to plan and adjust your workforce to the now ever-changing needs that companies are going to have, not just on the project level, but on a broader level of a company planning perspective, is becoming a critical thing that every business will have to learn how to do to compete in a world where AI is infused into everything. And now the good news is that AI can dramatically help you improve and get better and more efficient in doing that in workforce planning and understanding the requirements, understanding the resources you have and using them. And by using the right tools and the right processes and a few simple prompts, you can take a process that used to take teams of people days and do this with one person in minutes.

2:10And this is exactly what we're going to show you today. Now, the best person that I know to show this to you is Josh Houliere, who happens to be our guest today. And why do I think he's the best person to do that? First of all, he spent many years as a financial analyst and then as a planner helping businesses and corporations in exactly these kind of things. He spent the last five years in Ernst & Young helping businesses through business transformation and viewing it from a financial and resource side. So he has seen this both from within companies and as external advisor. And now recently he moved to doing his own thing, helping businesses through this transitional process with AI.

2:51So it really has a very broad and deep view of how to do this right. And as I mentioned, since this will become a part of every business moving forward, and maybe not in the next six months, but definitely in the next 18 to 36 months, all businesses will be around figuring out how to embed AI and how to run projects with AI embedded in them, and how does that impact the resources that are required. And I'm really excited to have him as a guest today. So Josh, welcome to the Leveraging AI podcast.

3:23In the next few years, AI technology will change our world dramatically. Whether you are a business executive trying to catapult your business forward, or just somebody who refuses to be left behind and want to advance your career, this is the show for you. I'm your host Isar Maitis, a serial entrepreneur and an AI enthusiast. You'll hear invaluable practical tips from innovative business leaders, AI practitioners, and some of the brightest AI minds in our world today on how you can leverage AI in ethical ways to advance your career and grow your business.

4:03Awesome. Yeah, thank you for having me. Hopefully everybody can hear me. Got a good audio check earlier. But yeah, so I want to basically cover how we can go about doing a project plan for workforce planning, specifically using AI. So I'm going to do a quick screen share and get that set up here in just a second. Yeah. While you're doing that, those of us who are joining us live, feel free to say hello in the chat, say where you're from. If you're not in here live, why aren't you here live? We're doing this every Thursday at noon PM Eastern, and you can come and join us and ask questions and be engaged in the chat.

4:34It's always very active. One last thing that I will say before we get started, Josh, this show is sponsored by the AI Business Transformation Course. The AI Business Transformation Course, I've been teaching it personally since April of 2023. And since then, I've taught probably thousands of business leaders on how to transform their businesses using AI. The course covers an introduction, so those of you who know nothing can get a lot out of it. But also it teaches a lot of multiple use cases across different tools and how to use them and how to apply them for business. And it ends with an actual blueprint that I use with my consulting clients on how to implement AI from a company-wide perspective.

5:14The course is four sessions, two hours per week. And we just opened our next cohort. It's starting on February 17th. the number one thing that's going to decide whether you're going to have successful AI transformation or not is training, education, and skills for employees and leadership. That has been proven through multiple research and surveys and whatever. It's the number one decision. If you don't have a means to train yourself or your team right now, don't miss this opportunity. I teach this course all the time, but most of these courses are private, meaning I teach specific organizations who invite me to teach them.

5:54And I do a public course like this once every quarter or so. So don't wait till May of 2025 to get your act together when it comes to training yourself and or your team with how to implement AI in a business. Come join us February 17th. There's going to be a link in the show notes so you can go from there straight to the signup page. But with that, back to the very, very important topic of today. Oh, one thing that I forgot, since you're listeners of this podcast, you can use the promo code leveraging AI100. So basically the name of the podcast, 100, to get a hundred hours off the course. But now back to the very important topic that we're going to talk about today, which is how to plan for projects and how to do workforce planning using AI in the process.

6:39Josh, back to you. Absolutely. Yeah, appreciate it. And yeah, great course. It's a four-week course. I think it's like every Monday you're hosting it. So definitely a great crash course for businesses that want to learn how to use AI and adopt it at their organization. So I highly recommend it. So with that, there's really three things I want the group to walk away from today. First, what is workforce planning? What does that mean? What is perplexing AI and how can you use it in terms of workforce planning? And then really, what are some prompts that you can use today to actually start creating a project plan for your workforce planning implementation?

7:09So a couple of things here. So in terms of what is workforce planning, it's basically a method that companies use to budget and forecast employee related expense. So think of it in terms of like headcount, employee salary, employee benefits, and things like that. So a really good example would be, let's say that your company has 10 ,000 employees. And let's say that you have a manager that has three different departments. And in one of those departments, they have 10 employees. So what they need to do is be able to forecast, currently, I have 10 headcount, how many will I have for the next year?

7:39So maybe they want to do a new hire, maybe they want to hire two new junior resources. And then maybe they want to transfer one of those employees from say like a sales department over to a marketing department. So that'd be an example of how they want to do planning from a headcount basis. On top of that, you also have people's employee salary expense. So the other thing that you want to try to account for is what's my annual merit increase. So in terms of like merit increase, we're all familiar with inflation. So inflation is high. So usually what merit increases are going to be doing is one, recognize people for performance, but also try to make sure that you're keeping up salary in relation to inflation to some degree.

8:12So the industry average is usually an increase of about two to 3%. And then the last thing is you want to make sure that you have an annual bonus payout. So a lot of companies will have a performance metric that employees and the company need to attain. And then upon attainment, basically pay out a bonus off of that. And then of course, there's going to be things like benefits. So think of terms of 401k, health insurance plans, dental plans, things like that. So that kind of encapsulates what's all included when you're planning for workforce planning. Now, in terms of perplexity - And again, you know what?

8:39I'll pause you just for one second. As I said in the beginning, this is going to become so dramatically different as far as the pace this needs to happen. Because so far, this was something we've done for years, right? It's not new. But the changes that AI itself will introduce, the amount of curveballs that are going to be thrown into this process, will have to dramatically change the way this happens because we may not be able to do this just once a year. We may need to do this once a quarter or once every six months to adapt to the change that AI will allow us to do. And if we don't, we may be less competitive.

9:16And it's on both aspects of the spectrum. On one hand, it might be a bloodbath as far as going after talent with AI knowledge in the next few years until everybody catches up. And on the other hand, you may not need as many people on some departments, but more people on other departments. So the structure of the workforce may shift dramatically, which will require this process to be even more efficient than we can do it today.

9:43Absolutely. For the next part, what we wanted to cover was basically how do you use AI to then help you do workforce planning? So in this case, what we're going to use is perplexity AI. A lot of people are very familiar with ChatGPT. And so really, what's the difference between using, say, ChatGPT versus perplexity AI? Well, a lot of it has very similar characteristics in terms of it's a large language model. You can use natural language to basically type to it, ask it questions, have it perform tasks and things like that. When you do have a pro plan with Perplexia AI, you're actually able to use multiple AI models.

10:15So currently you're able to use ChatGPT from OpenAI, Claude from Anthropic, one of the main competitors to OpenAI and ChatGPT. And three, the new DeepSeq model, which is the crazy model that came from China that everybody's been talking about. So if you haven't heard, DeepSeek recently released a new model, and it caused quite a stir in the stock market earlier this week. It actually caused NVIDIA, one of the main AI companies out there, to have a very sizable drop in their stock price because of the impact that this model has on the broader market. Only half a trillion. It wasn't a big... Yeah, only half a trillion.

10:48Just a little bit of a big deal. A bump in the road. Exactly. And so some of the other benefits that you have are perplexity AI is when you ask it a question or you have it do a task, it's actually going to look up multiple different sources. It'll give you the ability to do a dropdown and see where those sources are being pulled from so that if you wanted to vet or just validate where that information is coming from, you can actually see what specific sources it's pulling from and actually read through that in more detail. And then in relation to that, it's providing citations and references.

11:16So as you'll see, there'll be like little citation marks. So if you're saying like, oh, where did it get that number from? Or where did it get that information from? You can click the little citation and then it'll take you over to the webpage or the article where it pulled that information from. So just to pause you here, first of all, I love perplexity. I lose perplexity way more than I use Google in the past, I don't know, six months. But just so that other people know of other options on tools that do similar things. First of all, OpenAI is introducing that into their models as well, whether it's going to search the web and do things.

11:45I still think between ChatGPT and perplexity, I like perplexity's outputs better than I like OpenAI's. But there's two other tools that are out there that do a similar job, only better than perplexity. One is not free, which is Google, Gemini, Pro, DeepSearch. And so if you have the paid version of Gemini, it does something very similar, only it does it better than perplexity. It will go through a lot more websites. So perplexity usually will go between 8 and 20 different websites. The Gemini model will go between 80 to 250 websites to look for information for you. And it also looks for more. Both these models have this agentic behavior where they will define for themselves what sub things to look for.

12:29So it tries to understand what you're looking for and will run several different research patterns to find relevant information. Google Deep Search will do it better than perplexity. And I think over time, Google will win this just because they're Google, right? So from researching the web, they just have an advantage that nobody else has. The other one that is free is actually called you.com, Y-O-U.com. And it's very similar in its capabilities to Google. And it does a really great job, but then you're going to be limited. So if you want to run 20 of these a day, you won't be able to. But if you want to run two or three a day, it works amazingly well.

13:04So that's two cents to other options other than perplexity. But that being said, Perplexity is awesome. And the free version is good enough. Absolutely. Yeah, that's a great thing with all these AI models. It's good to have a variety in the marketplace that keeps other models in competition to have more and more development and capabilities, as well as helping drive the cost down in the long run so that users like everyone on this call can more easily afford these types of tools. So in terms of using Perplexity AI for this particular use case, which is what we want to do is create a project plan for workforce planning that we can actually implement into a solution.

13:38So when it comes to like workforce planning, again, what we're trying to do, trying to forecast like how many employees do we have like in a department for next year? So typically what you can do it in Excel, something quick and dirty, a lot of companies will do that. More often than not, as your company grows and you start getting to more of a scaling size or more of a global company, it's better to have really like an enterprise performance management type tool. So a lot of the common tools out there are going to be Oracle, SAP, OneStream, Adaptive, and Anaplan. So for this particular example for today, what we're going to try to do is have Perplexity specifically give us a project plan for implementing workforce planning in a cloud-based planning tool called Anaplan.

14:16And so with that, what you'll see is basically three simple prompts that we're going to put into Perplexity AI. And what these prompts are doing is really three parts. So we're going to be creating the requirements, we're going to convert those requirements into user stories. And then we're also going to then take those and plan our actual project. A very quick overview in terms of what Anaplan is. Anaplan is a cloud-based planning tool used by many Fortune 500 companies. So some popular ones are NVIDIA, LinkedIn, Google, AWS, Starbucks, and more. So it's a very robust tool. It's been around for well over a decade.

14:50And it's actually one of the tools that I've been implementing for the last decade for a lot of our Fortune 500 clients in the past. And then the last aspect of this, before we jump into the live demo is we want to understand what is agile. So when we think about trying to do a project plan and implement something like workforce planning in a platform like Anaplan, we need to have a framework or methodology for how we're actually going to go about implementing that. In terms of project management, what we're doing is actually using something called the agile implementation methodology. And really that's just a very structured format in terms of gathering requirements.

15:21And then what we do is you try to turn those requirements into what are called user stories. So what you try to do, you try to identify here are the different roles within my organization that need to be able to do planning or analysis or reporting. And then from there, you take those user stories, and then you basically allocate them into something called sprints. And really a sprint is just the amount of time that you're allocating to actually do the implementation and the development of the tool itself. So oftentimes a sprint will be anywhere from two to three weeks. And some of these projects, if it's a smaller use case, you can get it done in say three months.

15:51for some of the larger projects like NVIDIA, for example, they're actually using Anaplan for their supply chain planning. So in terms of all of the chips and the GPUs and things like that, they're actually using Anaplan to forecast that in this cloud-based planning tool. And they've been doing that over the last four years. So again, some of these use cases, you can get them quick and dirty three months in and out. Some of them, they're truly transformational in terms of multiple years. So before I jump into the demo, Isar, anything that you want to cover? Two quick things. First of all, the concept of agile methodology and user stories and sprints started in the software world many years ago and over time adopted by many other industries.

16:32But the idea is very simple. The idea is saying, I want to maximize the resources of my organization. And if I plan a four-year project, a lot of things are going to be weird in between and are going to change and so on. And if I can break this down into the smallest requirements possible, meaning the actual task level, I can plan much better. And the planning literally happens on those sprints. And every company or every department does it a little differently. So some companies do sprints of a week, some two weeks, some three weeks. I don't think I've seen anybody do more than three weeks. Most companies do it around two-week sprints.

17:04And within that timeframe, you literally plan down to the task level for every single individual involved. And there's multiple types of software that allows you to A, assign the task to people and B, track the completion of these tasks. So it allows you to, on every day, on every week and every sprint, see exactly where you're standing in the very detailed progress. So it's not like a wishy-washy somebody in your organization saying, well, we think we've done 30 % of the work, which is what used to happen when we used to run these Gantt chart gigantic projects. And then Agile really allows you to see exactly where you are.

17:37So that's the benefit. And this is like what we're going to talk about. Absolutely. Again, if there's any questions from the group, feel free to drop them in the group chat. And so with that, we're going to jump into the live demo. So first, a lay of the land, what we have is perplexity AI. And this is basically the interface. So you basically go to perplexity.ai, you log in and create an account for the first time. And essentially, what you're going to have is on the left here, a side panel, where you basically start a new thread. And the new thread is basically where you enter your question or your prompt or the task that you want the AI to achieve.

18:08And then there's a couple of other features that we could highlight, but we're going to focus primarily just on the thread. And so essentially what you're about to see is I've got three template, three prompts that are basically, I'm just going to copy and paste. And then we're going to look at the input and then talk about the different type of aspects that Perplexi offers from these prompts. So first things first, in terms of this first prompt, essentially what it's just saying is what I want to do is have an expert in workforce planning and agile implementation to basically craft requirements for companies to budget and forecast their headcount.

18:40So a couple of things in terms of just like prompt structure before we jump into running this through perplexity. When you're trying to craft a prompt, there's a specific kind of set of guidelines that you want to do. You want to basically assign it a role. You want to give it some background context and then be very discreet or very concise in terms of what specific task you want it to perform. So I try to liken it, if you imagine ChatGPT or perplexity or all these other AI models, they're trained on the entirety of the internet. So they have very general knowledge. And so if you imagine the entirety of the internet is like the sun, right?

19:10The sun gives a lot of light. What you're trying to do, you're trying to take that sun and focus it down to a spotlight and then focus that down to a flashlight and then ultimately focus it down to the laser pointer. Because what happens is the more precise you are with your prompts, the better the output you get. If you're going to be very generic and general with it, you're going to get a very generic and general output. And so what this is doing is trying to structure it so that you have that focus laser pointer. So an example would be like as an expert in workforce planning, right? So rather than trying to cover everything within a company, sales, marketing, finance, operations, focus it in just on workforce planning.

19:44And then second, an expert in agile. So it could use all sorts of implementation methodologies, but we wanted to specifically focus on agile. So a couple of examples as to like how you want to basically use your prompt and take it from everything under the sun down to a laser pointer. So with that, I'm going to go ahead and copy and paste this prompt. Go to perplexity, drop it in here. And what you can see here is I have the pro plan. So with the pro plan, I could basically use reasoning with the new deep seek. I could use chat GPT or the default that I have set up behind the scenes is basically clawed from anthropic.

20:17So I'm going to go ahead and hit run. And as you can see right now, it's using some of the reasoning logic that a lot of these different models are using. So it's saying that it's going to research for best practices for workforce planning, identify any core calculations, and then start gathering requirements for any best practices for dashboard reporting and analysis. And as you can see here, it's so far gathered about 24 sources. And then from here, you have a little preview of what those different sources are. So if I wanted to, I can click on show all, and I can actually see in detail, these are the different 24 sources that are being pulled for this response.

20:53Now let's take a look at what it came up with. So essentially, yeah, for headcount planning, that's basically saying that there's baseline requirements. So basically wants to have to track your existing headcount. In our example earlier, we've got a department with 10 existing headcount. Then we also want to be able to forecast any new hires and then even do any sort of transfers in between departments. Say I want to transfer someone from sales over to marketing. And then I'll go through more detail around the calculation components. It'll focus in on how you go about calculating salary. More importantly, how you calculate that merit increase.

21:21So you want to reward your top performers with merit increases on an annual basis, keep them motivated and reward them for good work. And then likewise, anything with bonus target planning and setting. And then, of course, the benefits packages. So things like 401k, health insurance, dental insurance. Then this next section is going to go through the dashboard requirements. So, again, a lot of these applications could be in Excel, could be Oracle, SAP, in this case, and plan. What you want to do is you want to have a user interface. Right. So how am I actually going to go in and plan this headcount?

21:52How am I going to view how many people I have in a department? How am I going to be able to do scenario analysis from forecast to forecast? So these are the different requirements in terms of those dashboards where the user actually interfaces. And as I mentioned before, there's these different little drop downs and things like that. And so what you have is like a citation here. So what I can do is say, where did that come from? And then I could basically go here. And this is where it's pulling that information for how to come up with that dashboard as an example. So then going back down here, it goes to the whole thing.

22:22So you can see how it's very detailed. Again, that's because we took our prompt from being very generic and try to focus it in like a laser pointer. So suffice it to say, we've got all of the requirements there. So in terms of the next prompt, what we want to do, we want to basically break... Before you dive to the next prompt, I want to touch on a few very important things that kind of are in behind the scenes of the stuff that you're doing. The first thing is the importance of a prompt library, right? So in your organization, instead of every person trying to do the process that they need to do on their own and try to figure it out every time they're doing it again and again, you want to hold a centralized prompt library, which is basically what Josh is doing.