Leveraging AI for GovCon Success

In this episode of Game Changers for Government Contractors, host Michael LeJeune explores how government contractors can leverage AI to improve efficiency, streamline processes, and enhance proposal development. Drawing from real-world examples, Michael shares insights into the key distinctions between AI and automation, and how AI can be used for red-teaming proposals, transcribing documents, conducting data analysis, and more. He also provides practical tips for creating an AI policy, managing security concerns, and getting the most out of AI tools. Whether you’re new to AI or looking to deepen your understanding, this episode offers valuable advice on how to make AI work for your GovCon business.

To listen on your favorite podcast platform, click below:

???FA: https://federal-access.com/ep-341-leveraging-ai-for-government-contracting/

???Apple Podcasts:?https://apple.co/3y4sNdA

???Spotify:?https://spoti.fi/3SPTZoB

??To watch, click here: https://youtu.be/nLXWElyxAvI

?

Read Transcript Here:

Michael LeJeune (00:00): Welcome to another episode of Game Changers for Government Contractors. I am Michael LeJeune, your host, and today I will be talking about leveraging AI for government contractors. I'm going to hit a handful of things. This is a session that I've been doing all around the country. People are really excited to learn about AI and how they can more effectively apply this in the government contracting world.

But before I jump into that, let me tell you real quick a little bit about RSM Federal. If you are not familiar with RSM Federal, we've been in this space for about 15 years now, helping government contractors win a little over $14.5 billion. We've taught thousands of companies around the world how to do this, literally in plenty of different countries. We have five Amazon bestselling books out there. My latest one is called I'm New to Government Contracting. Where Should I Start? So if you are new to government contracting, I highly recommend you go pick up that book on Amazon today. Even if you've been in the market for 5 or 10 years, I would highly recommend you go pick up that book; you will learn a lot from it.

But with that, let me dive in and talk about AI and government contracting. I want to set a little bit of a foundation and give you some practical examples. If you're watching the video podcast version of this today, know that I'm going to be looking off-screen quite a bit at my notes because I want to make sure I cover everything—because there's a lot to cover. And by the way, if you are listening on audio and would like to watch on video, this is on YouTube. Look up my name, Michael LeJeune, and you will find the YouTube version of this that we stream every week at the same time that we stream out the audio podcast.

So go do that. And as a bonus, if you also like to read or prefer to read it, our newsletter on LinkedIn is a way you can go and read it. Every Wednesday, we put it out as a newsletter on LinkedIn. So if you go onto my LinkedIn profile, you will be able to find how to subscribe to the newsletter so that you will get the transcript of the podcast in there with all the links of where to watch and listen.

So with that, let's jump in. In my opinion, I think we have gotten way out of control with AI as far as overthinking and overanalyzing what it is, what it's going to do to the market, and what it's going to do to jobs. All these different things that people are freaking out about. So hopefully, in the foundation here, I can help you understand that this is not the big bad wolf that a lot of people are making it out to be. I just don't believe that. And with that, I'll start with this: It's garbage in, garbage out. If you're not very good at what you do, AI is going to make you a little bit better. If you're excellent at what you do, it's going to make you outstanding and allow you to speed up what you're doing. So it really does create a level playing field for the users of AI.

I think if you are sitting on the bench and are afraid of AI, that's the problem. Those are the people who are really going to struggle—the ones who refuse to adopt it because they think it's evil or that it's going to be watching everything and taking notes on you. I mean, the government already takes notes on you—not to surprise you anyway. But it's not the evil thing that a lot of people make it out to be. It doesn't create an unfair advantage; it creates a level playing field for a lot of people. So it's garbage in, garbage out. If you're already good, it's just going to make you better.

Another tip here from my perspective—and I talk with my partner Josh Frank about this a lot of times—we both agree on this: Processes should drive the technology choices you use, not the other way around. Just because ChatGPT is available doesn't mean you should use it. What would you use it for? How would you do it? Your processes should be driving the technology choices you make, not saying, "There's a new iPhone; there's a new camera; there's a new laptop; there's a new version of this software, and I need it. I must have it." Your process should determine that, not the popularity of the tool.

I want to make a distinction here before I go any further between AI and automation. Those are two different things. AI creates and learns; automation performs tasks. Now, automation tools can interact with AI to create and perform tasks. It's very interesting what people are doing. Zapier is one of the primary tools that a lot of people use and connect with ChatGPT and other AIs where they will have a series of tasks that they want to perform, and they will build a workflow inside of those tools that interacts with AI, grabs information, does certain things, and AI populates the tool, the workflow, the process—whatever you want to call it—and then it continues to execute on that. Monday.com and a lot of other CRMs have workflows built into them. These are situations where the system is saying, "If this happens, do this task. If this happens, do this other task. If this doesn't happen, stop the cycle or do something different."

Automation happens in just about everything from CRMs and program management tools to people using it in their sales process a lot, and in things like Canva as well, where you can say, "Next time I'm doing something, it needs to create these images," or whatever it may be. Or people use it in Slack. So if I get a Slack message, it's supposed to do this or that. There are so many of these tools out there. Just understand there's a difference between AI and automation. The two work together, but they're actually two different things.

So with a little bit of that foundation out there, I want to talk about some practical examples that our clients are using this for. I'm going to talk about policy creation and ethical considerations and all that stuff toward the end. But I'm going to say this: Understand that if you're scared of AI, if you have security concerns about AI, if you have any other concerns about AI—either don't use it or research it. I had someone recently in a session that I gave on this example I'm about to give you who totally had a meltdown. They literally, their brain melted when I put this out there because they're like, "My God, there are so many legal considerations, and this, that, and the other." I didn't say you had to use it. I didn't beat you with a stick or hold a gun to your head and say you've got to use AI. If you don't want to use it, don't use it. If you think there's a problem, don't use it. However, people are using it. They're creating policies, and they're creating ways to protect data and overcome security concerns. So it can be done. Go into this with an open mind. That's a big deal.

So let's talk about red-teaming proposals. I have clients that are using AI to red-team proposals. If you've ever red-teamed a proposal—this is a review of a proposal—you know it takes hours and hours and hours to red-team a proposal. But what if you could upload the RFP, upload your response, and have it do the red team for you and make suggestions to your document? What if—because people are doing it right now—and then they are taking that red team as the first sweep through and then doing a final sweep with human eyes just to make sure it didn't miss anything? So imagine getting all that feedback. Because here's the thing people don't think about when it comes to AI: AI is going to look at everything. It's going to follow the commands you give it. That's the big thing. It's going to follow the commands. I highly recommend you make the commands simple. I'll talk a little bit about that here in a minute.

The human eye and the human brain do not see everything. Have you ever walked into a room and said, "Where are my glasses?" and they were on your head? Or, "Where are my glasses?" and they were sitting on the counter, and you didn't see them, and you looked all around? And finally, on the second, third, or fifth trip into the kitchen or living room, you're like, "They've been right there this whole time." The human eye and the human brain don't catch everything. AI is going to catch everything—or at least all the important stuff that you tell it to.

So I highly recommend using it to do things like red teams or review different things because it's so valuable in catching things that you may not catch. Even if you have red-teamed a hundred proposals in the last year—which, God bless you if you've done that—that's a lot of work. But even if you've done that, you can get in a rut. You can get tunnel vision. You can get night blindness—whatever you want to call it—and miss something. In fact, here's a fun little fact for you: Did you know that most accidents happen within the last seven miles between someone's home and where they are? Because we often get this driving blindness where your brain goes on autopilot and is just not paying attention. And I can confirm that by just saying, "Have you ever gotten home and not realized how you got there?" because your brain was on autopilot. The same thing happens when it comes to your work and reviewing a proposal. You will get "proposal blindness"—let's use that term. I'm coining that one today. And proposal blindness will make you miss stuff. So use the AI for that.

Now, here was one of the concerns people have, like, "What if there are CMMC implications? What about sensitive data?" Again, go into this with an optimistic and open mindset of just redacting that information. Just redact things that are sensitive. That's all you have to do. That's it. We'll talk about policy in a little bit.

The next example is that I have a client that does language services, and they transcribe a ton of documents. This was all done by hand for many, many years. Often, it's complicated, and it's up to interpretation by the transcriber. And so there are consistency issues and all different things like that. What they're now doing is using AI to try to make the first transcript for documents, podcasts, videos—whatever it is—and then having the experts review it. So what they've done is they have reduced their costs by 90% in most cases—sometimes just 80%, but 80 to 90% reduction in cost because they're having a tool doing the first swipe and then a person reviewing it.

So what does that mean? Because I know somebody's listening to this saying, "So now you're telling me it's going to reduce the wages on that person because now they're not getting enough work?" What it's allowed the company to do is scale up the amount of work they're doing and give that person more work or at least the same amount of work. They're doing more jobs for them, making them more valuable to the company. So no, those people didn't lose income. The company increased revenue and profits significantly, allowing them to work with those people more.

So that was the second one—language services, but transcription services in general. The third one kind of ties into that. AI can transcribe anything. So you can take a photo, anything you can think of—business cards, a whiteboard, a brainstorming session, large printed data sets. A friend of mine was doing something the other day, and there was a roster sign-in. It was one of those things where, like, I don't know if you've ever been to a conference, but when you go to a conference, everybody signs in, usually. So, name, phone number, email address, sometimes your small business status—whatever it is.

So think about this large data set of printed information that you would normally get back to your office and hand to the secretary or somebody and say, "Hey, could you please type all this up?" It takes hours and hours, especially if you're at a conference and there are 3,000 people, and they've all hand-signed something. It takes hours to do that. You can literally take photos, upload them, and say, "Put this in an Excel spreadsheet," and it'll do it in seconds.

A popular thing as well is that people collect all these business cards, and they get back, and somebody's typing it into the system. It takes an hour or two to type all that stuff. You can literally just take a screenshot of blocks of them on your desk—not even individually. Upload that into the system and say, "Transcribe all this." So the transcribing power—it's just a time-saving thing that will help you immensely. So use the transcribing features in these tools.

The next one is analysis. It is so good at taking large data sets and comparing and looking for trends, whether it's pricing sheets or any kind of data. I will sometimes take social media statistics and data and things and put them in there and say, "Look for trends. Just give me an idea." It'll show the title of something, a paragraph about it, what the thumbnail was, and then it'll also put in there how many likes, shares, comments, that kind of thing that it got, how many views, the duration, where people are dropping out. There are so many good analytics in YouTube for stuff like this. And you can drop that in there and say, "Take a look at all these videos, compare them, and then look for trends—what is working and what's not, what type of headlines, what type of thumbnails," all those types of things you can figure out.

So that's just on this one side, but then you can do it on content as well. So what are the factors making something effective? You can take somebody else's content—like a competitor's content—and be like, "These people are killing it on LinkedIn. They're getting all kinds of comments and traction and stuff like that," or even their website, and say, "What is so good about their website?" You can upload those things and have it do some analysis for you and tell you what's really working. The limits on the analysis side of this are up to your imagination. It's really just your imagination—what do you want to compare?

So I would think about it from this perspective: What is something that you manually do today that you find very time-consuming but very process-driven? Like, I do step one, two, three, four, five, six—whatever it is. It's time-consuming, and it really doesn't take that much thought; it just takes effort and energy. That's a perfect thing to look at using AI for. Use the AI to do those routine, very simple tasks for you.

Often, for me, regardless of what I get from it, I'll either ask it more questions, or I'll wind up massaging whatever it gives me, especially in content and stuff like that. We'll talk about that in a little bit. But if there's analysis you need to do, AI is really great for that stuff.

So let me talk about policy creation for a minute. Your company needs an AI policy of some sort. Can you use it? Can't you use it? What can you use it for? My only recommendation is don't overcomplicate this. There's absolutely no reason for you to overcomplicate it with pages and pages of AI policy. If you're a bigger company, it's going to be complicated anyway because bigger companies are just complicated. That's the way they work. Try not to overcomplicate this. Think about data protection and security concerns. How can you mitigate those issues if you have them? Do you need to redact certain things? What types of things need to be redacted? How does that need to look in your business?

There are some ethical considerations around claiming that you did things. I don't claim I did anything because I actually do it. I look at the AI side of this as it's often a writing partner because it's extremely rare—it probably never happens—for me to take something that AI did and post it exactly the same way. But it's also like you're not claiming responsibility for it either. So just take that with a grain of salt. People are like, "I don't know." You're using a tool to do something. I mean, have you ever used Photoshop to create an image? The tool did it. Did you give Adobe Photoshop the credit for doing it? Probably not. I'm recording this on a Canon camera, and yes, I just gave Canon the kudos for that and the credit. But most people don't even think about those kinds of things. You're using a tool to do a job. If you typed up something on your computer and printed it out, you're probably not going to give Microsoft Word and your Canon copier credit in your document. It's a tool. Use it as a tool.

The last one here is training in the policy creation side. You should have something built into your policy about how people need to be trained, what training should occur in order to use it, and so that they actually understand what the policies are. The majority of the training is built on your policies and usage, things like that, and a little bit of training on actually using the tool.

So with that, let me jump into what I currently use AI for on a very regular basis. I use it for social media ideas and posts. I'm constantly asking it questions, and it gives me tons of ideas, or I'll run an idea by it and say, "Maybe wordsmith this," or whatever. But I use it for social media. I use it for reviewing corporate overviews because a lot of times, somebody will send me a bunch of information about who they are and what they do. Instead of me writing the first draft myself, I will ask the AI to take the information about the company and write a three-to-five-sentence corporate overview. And then some people will send me their corporate overviews, and I'll say, "Hey, here's kind of what I'm looking for in the corporate overview. Here are some metrics I want to add in and massage into this. Take a first shot at this," and the AI will do it. And then I will massage it myself from there. I've found it does a really good job from there.

The next one is documentation, especially when you have something that's audio or video. It can create documentation for you. It can do transcription. There are so many cool things it can do. Or you can say, "Hey, I need a policy for this. I need an outline for something. I need to document how to change your username and password on the website." And it will literally start that documentation for you so you have a foundation to work from. And that's what I love about AI—it just gives you these foundations to work on that just excel.

Because I don't know about you, but a lot of times I'll sit down to work on a project, and just like the email and password thing, I'll think, "Okay, what do I need to include in this?" And there's a lot of thinking that goes into it. Then later, I'll be like, "I should have added that. I should have added this." With AI, asking it to jumpstart the process often adds things that you wouldn't have thought about. And that's why I really love it.

I've used AI to write emails where I'm trying to say something, but I'm like, "I just don't like the words that are coming out." I'll say, "This is what I'm trying to convey. Please write an email to this person explaining this," and it'll just spit out the paragraph email. I'll make my modifications and then respond. Again, I don't use it all the time, but it's a tool when you are hitting a roadblock that's very easy to implement so that you can get through the roadblock faster.

You can create reports with it. You can create training material. I've talked a little bit about data analysis. Blog posts and articles are really good—again, just jump-starting the process for you. Something I use it for all the time is filling out complicated forms. If you have ever tried to submit a form to speak at a conference or there are so many different places where you go and there's a long form and you only have so many characters or so many words that you're allowed to use. Like, you know, my personal bio is 350 words, and it's limiting me to 100. I'll drop it in AI and say, "Hey, take this current bio and squeeze in all the important facts into 100 words," or whatever it is.

That's a really great way versus me pulling up a Word document and fiddling around with it and trying to figure it out. I'll just let the AI do that, and then I'll massage it. I've talked a little bit about creating outlines. I love to create outlines with AI. I've talked about transcripts, and then the final one is research. I do so much research with AI, and the reason I love it—whether you've got the Google Labs piece in your Google search or whether you're using ChatGPT or another tool—the thing I love about the research is it skips the ads. So instead of me only seeing all the ad-driven content—which, depending on what you're searching for, the first page or two pages might be ad-driven content, and that sucks—I don't want to sift through all that. I just want to see other things. And so that's why I like doing research with AI. It's a really good research partner.

I'll wrap up today by talking about the best practices. In my opinion—and this is Mike's opinion, take it or leave it—I feel like it's best to get the paid version of whatever you're using. If you're using ChatGPT, get the paid version. You have access to things that the free version doesn't have. The free version is outstanding, but I've found it to be night and day between the results I get from the free versus the paid version. So I'm an advocate of the paid version.

The second one is, all these AIs have some form of chats that you have going on. Name those very clearly so that you can go back to them. Because that's the next tip—go back to those chats to maintain the context. A lot of times you'll start a new chat. You'll start asking questions, and you have to kind of massage your prompts to get to where it's working the way you want. Then what happens is people will go and start a new chat, and it doesn't have the chat history, so it doesn't know. And so you've got to start all over with training the AI about what you're looking for. If you go back to the chats that you've named clearly, it maintains the context like you never left. Remember, it's a robot—it's like you never left.

And one other little tip here is when you are working with the AI, talk to it like a person. Compliment it, tell it things you like, tell it things you don't like. Those are things that will allow the chat to adjust to what you are doing, and it will get you better results.

One of the other tips here is, and I've talked about this so it shouldn't be a surprise here, but in the paid version of most of them, you can upload documents. That's one of the big differences that I really like about that. So I upload documents instead of copying and pasting, instead of having it go read a website. I upload documents to it and have it read the document and then do the work based on that. So that's a really good best practice.

And then I would say not overcomplicating your prompts but being specific about them is very important. Don't necessarily sit there and write a paragraph of prompts without breaking them out. What I usually tell the AI to do is, let's go with a document. I've uploaded a document that is a transcription of my podcast. I'd like you to perform two tasks. Task number one is to read the document. Task number two is to build a description of this that I can use on social media. I may even throw in a third task where task number three would be, "Give me 7 to 10 suggestions for the title based on what the current title is."

If you are very specific about the tasks you want to perform, it will perform all three of those tasks, and it will break out the tasks. Whereas if you just write a paragraph like you're talking to somebody, it may choose what to do or think it's accomplishing everything, and it may miss something because of the way you wrote it up. So that's why I try to be very specific with my tasks. I know this is what I need to get out of this.

So with that, I hope you enjoyed the session today. There's so much beyond what I covered. This is really scratching the surface on tips for using AI. If you have questions, please reach out. I'd be happy to help you. We'll see you in the next episode.

?