What You Need to Know About AI to Use it Effectively

I’m about to read some AI terms.

I would love to know in the comments if you know these.

Back propagation, K-means clustering, hyperparameter tuning, feature engineering, cross validation.

Now the question is, do you need to know these to utilize AI effectively in your business?

I hope not, I sure don’t.

If you’re an executive leading an AI transformation in your business, these probably aren’t the things you need to know either.

But today, we’re going to talk about what you do need to know.

Hi, I’m Courtney Baker and this is AI Knowhow from Knownwell, helping you reimagine your business in the AI era.

As always, I’m joined by Knownwell CEO, David DeWolf, Chief Product Officer and Chief Technology Officer, Mohan Rao and Chief Strategy Officer, Pete Buer.

We also have a discussion with Aby Varma about what executives need to know about AI native products.

But first, let’s turn it over to David and Mohan for a discussion around what you need to know about AI to use it in your business.

Hey, David.

How are you, Trey?

Hey, Mohan.

I am great.

We had so much fun alone last time.

We kicked Courtney out again.

Exactly.

We’re going to try to survive one more episode without her.

So, David, did you know that in the Google IO keynote back in, I don’t know when was it, last month, the term AI was used more than 120 times?

Whoa, did you actually go through and count those?

I think Gemini did that for us.

Was it hallucinating?

No, I’m not surprised, to be honest with you.

Yeah, it is a word.

Is it even a word?

It’s an acronym, but anyway, it’s used so much.

So David, let me ask you, what are the things that the business leaders need to know about AI to talk sort of intelligently, but also use it in their business?

Man, that is so hard because it’s so multi-dimensional, right?

To, you know, to, let’s just start with some of the basics, right?

I think a lot of business leaders equate AI with a very specific LLM and even more than the LLM, the user experience that they’re using to access the LLM.

So like they think AI equals ChatGPT, right?

I think having a basic understanding that, hey, there are a lot of different LLMs.

There’s kind of a race.

There’s a top tier amongst, you know, open AI and GPT and ChatGPT is obviously the way most people access that, but Claude and Gemini.

I think a lot of perplexity in what they’re doing.

Each one has pros and cons, and I don’t think we need to get into those, but I think a fundamental understanding that artificial intelligence is a lot more than a single LLM.

And then I’ll expand it even further.

All of these, I’m talking LLMs.

These are generative AI large language models that we’re talking about.

Artificial intelligence is really the ability to learn and apply knowledge.

And there’s multiple techniques, not just generative AI and LLMs that help with that, but also even beyond that, traditional machine learning is still very, very relevant.

And those types of things that I think having a good handle around just those fundamental concepts and not pretending like ChatGPT equals AI is something that’s just foundational that I still see leaders tripping amongst themselves.

That’s so true.

Also, I think there are several leaders who think of AI as entities as opposed to software and data systems.

So they’re just these prediction machines as opposed to this big black box that’s an entity that you really don’t know.

Almost a personality, right?

It’s almost like this big scary machine in the sky, right?

Exactly.

How would you kind of start with these LLMs?

You know, just let’s go with LLMs.

How would you start that in your company?

How would you integrate it with your existing systems?

What’s a good roadmap, David, to get started?

Yeah.

So I think first understanding what we said is tier one.

Tier number two to me is actually understanding some of the concepts of artificial intelligence.

So that definition of to know and apply knowledge.

But then how does it do it?

Where does it excel?

Here’s some really few just like conceptual things I find a lot of leaders don’t know.

Number one, they don’t know that what the difference is between what these large language models can do in terms of inference, right?

Which is probabilistic versus the deterministic math of previous types of compute, right?

There’s a difference between adding two numbers together or algorithmic computing and inference and how we’re able to probabilistically look at things and predict what a human would do next, right?

Those two things are very, very different.

And I think leaders need to have an understanding of that.

Okay.

Second level is then looking at, okay, based on that, what is AI good at?

Where can I apply conceptually this tool to real problems?

And to me, you can really think about it basically as it’s great at connecting dots, finding disparate points of data that don’t seem connected, but are connected.

And then number two, finding signal in the noise, parsing through volumes and volumes and volumes and volumes of information, and finding what really matters in it.

To me, those two things are the best ways for leaders to say, okay, I have a handle around AI.

So I would go there next.

And then finally, I think you get down to the question you really asked, which is about how do you integrate this into your system?

I think this is personally, I think this is highly dependent on the organization.

I think we are in a point of time where absolutely some of the technology leaders will be implementing their own things and building.

I think there’s actually too much building going on right now.

I think way too many organizations are cleaning their data and doing these big modern data stack implementations and preparing the data for their own custom algorithms and blah, blah, blah, blah, blah.

When in fact, what you’re going to see is a wave of innovation of brand new products that are coming out, that are going to solve a lot of these problems, and what organizations are custom building or going to be in the next 6, 9, 12 months, just commercial purchases off the shelf.

I’d actually be careful about doing too much direct integration unless you’re a high tech firm, unless you’re a massive financial institution with a very, very unique data asset.

I’d be a little bit careful to be honest with you.

And I would start with the usage, not the integration into the back end.

I’m curious how you’d respond to that, though, because you often have different perspectives on some of these things.

You know, I think you have to go with wherever your major pinpoints are.

Right.

So the first question is, do you integrate it or do you use it standalone?

Do you do you buy versus do you build?

And what do you do for what?

So I always go down to what are the biggest pinpoints that you currently have?

How can AI help with that?

Secondly, how can you serve your clients better if you use this?

Right.

So I’m just going to start with those two framing statements and then say, there’s so many areas where AI can be used, whether it’s build or buy.

You could say, let me look at my customer management, sales and marketing.

This is all of the front office functions that are super important.

Is there a role there?

Is there a role in the supply chain management area?

Is there a role in software development, if that is a core capability of your company?

Can you use it in finance and accounting?

I mean, ultimately, it has to be about increasing productivity, doing things more efficiently, or being able to provide a new service to your customers that’s better, faster, along those dimensions.

So this is sort of how I would think of it.

So build versus buy, integrator or stand alone, and be able to come up with an AI strategy that makes sense for your company.

And I think the other thing beyond just driving efficiency and productivity that I think about is creating new value that wasn’t possible before.

How can we leverage AI to create a new data asset, a new insight, a new level of intelligence that we literally couldn’t come up with before because maybe there was too much data for a human to go through and do it or is way too cost prohibitive to be able to do it.

Does it unlock some things?

And I think oftentimes when people talk about AI, they’re actually thinking less about the tooling.

They’re thinking more about the breakthroughs.

They’re looking for the big, they know this is going to be transformative.

The reality that we have seen in the first, I don’t know, year and a half now, almost two years of democratization of this technology, is that the big change, the big lift is that we’re getting 5%, 10%, 30% incremental improvement in what we’re doing, right?

And I think it is sometimes the usage, not that big breakthrough, but it’s those little usages where we’re getting incremental value we couldn’t have gotten before and doing things in new ways and better ways that is really impacting every aspect of business, not just one big bang change somewhere.

David, let me ask you another question.

Let’s say you’re about a $50 to $100 million services company.

How should people think about budgets and timing?

And how should they integrate these things in?

Because we have a lot of empathy for business owners and entrepreneurs running these businesses.

It’s hard and you’re dealing with your own hurricanes at any given time.

How should they think of this in terms of budget, resources, timing and such?

You know, I think the biggest resource actually drain is actually not the money, right?

Especially in this world where I think more and more is going to be commercialized products off the shelf and it’s not big implementation.

You’re not training models yourself.

As a $50, $100 million business, you’re not going to go train your own model, right?

You just don’t have the resource.

You can’t.

And it’s probably not prudent anyways, even if you did.

It’s the energy and the time suck to do something new that I think can be the biggest expense.

This is not AI-specific recommendation, but this is just general advice.

I think organizations that size really stink typically at stopping things.

I think we do a lot of strategic planning where we talk about that next big thing we need to do.

I think we analyze what’s already in motion and how do we turn it off, right?

Focus is about what you say no to, not what you say yes to.

And it’s pretty easy to say no to something.

If I have to choose implementing one of 10 things, I can choose one and leave nine on the side.

But if I have to turn something off that’s already going, is that project that’s already underway?

Maybe it’s pull something that I said five minutes ago.

Maybe it’s my modern data stack implementation that everybody was doing in a fellow.

If you’re not getting value from it and you’re not seeing a path to value, I don’t care what the world is saying.

Do you really take a hard look and say, should I have all these data engineers?

Should I have all this investment going into building this modern data stack that hasn’t actually proven to give me anything?

Or do I need to stop that, step back, look at what you said, which is what are my real pain points?

What are the real things I need to solve?

Say, forget lost investment.

Put that aside.

I’m starting fresh today.

Where’s the best use of my capital?

Not just my financial capital, but my human capital, my time, my energy in terms of carrying the business forward.

I think if you can do that, you can start to see clearly, what should I turn off in the business that’s not delivering a return or at least not as big of a return that I can get from some of these other investments.

That makes sense.

You’ve got to make time for this.

It is about priorities and ambition.

Last question as we wrap this up.

Who would you interest this project to?

How would you get started?

If you’re the CEO, you could either create like a tiger team, you could create a cross-functional team, but you’ve got to put somebody in charge for things to move.

It probably varies from company to company, but what’s the profile of somebody who should be in charge of this?

I’m going to go back to what you said about finding pain.

I think one of the big mistakes organizations are making is that they’re making this a technology problem and it’s not a technology problem.

I believe very strongly that you need to have a business leader that really is responsible for that pain point and understands in depth the business problem that you’re trying to solve that is sponsoring and leading.

I think that person can be paired with a technologist, but I think that even more important than a handle around the data and the technology is actually really clearly defining what that use case is, what that pain point is, and scoping the business problem to solve, and making sure that the solution is driving and not the technology driving.

I would look into my business units.

Depending on how you’re organized, it could be a holistic business unit, it could be a department, but really look for those business users that are responsible for the business metrics, the business outcomes, and have some sort of stumbling block that’s getting in their way that you think can be solved with this technology.

And so start there, and then bring the technology in.

Fantastic.

David, I think that is a wrap.

Do you have any other last thoughts before?

Last thought, like a whole lot of fun.

We’ve got to kick Courtney out more often, because we can just go at it, but I can’t wait to have her back either.

Exactly.

We will come back soon.

Take care.

Thanks, Mohan.

Hey, guys.

Courtney here.

Y’all know I love AI.

I love talking about AI and learning about AI.

You know what else I love?

I love marketing.

It makes total sense, right?

I’m the CMO of an AI company.

And I’m really excited to invite all of you to have a discussion with me, Mohan and two fantastic guests, Dom Colasante from 2X and Jeff Livingston from Cognitive Path, about just that, about marketing agencies, client retention in the AI era.

We’re at such an interesting point in time.

And for all of the marketing leaders out there, agencies, we would love to have you on August 22nd.

So right now, go register for the event.

You can do that at knownwell.com/marketingroundtable.

Aby Varma is a global marketing leader known for driving innovation and growth with high impact marketing strategies and digital transformation initiatives.

He sat down with Pete Buer to talk about AI native products and what executives need to know about AI.

Aby, welcome.

It’s an honor to have you on the program.

Thank you very much for having me.

If we could, can we give listeners a little bit of background, context for the conversation.

Spark Novus and your mission, please.

Absolutely.

So we really empower marketing professionals with AI, enhancing efficiency and efficacy and marketing ability to drive growth.

And really, we do two things.

We do consulting and advisory services for AI adoption into marketing teams, making sure that they can think about strategy, governance, use cases, technology.

And then the other thing is we do AI infused marketing services, which is essentially, we use AI to deliver traditional marketing services, what one may get from an agency, but we are able to do it faster or at higher quality and more economically than what a traditional agency can do, because we leverage AI.

This includes things like messaging, branding, content, social digital operations and that sort of stuff.

And we are very transparent about the technology that we use and work with marketing teams to make it happen.

Marketing is one of the disciplines in the enterprise, where we actually see a fair amount of experimentation and use case.

I wonder from your perspective, you’ve worked with a bunch of companies and marketing leaders.

Where are the most promising use cases out of the gates in deploying AI?

I think there’s sort of two areas in my opinion.

So one is just purely from a content standpoint, right?

Is making sure that how can we deliver content, whether it’s at a very foundational level in terms of just messaging and positioning.

Obviously, that needs a human aspect.

But then leveraging all of that foundational work that you have done and converting that into a kind of a cohesive content marketing strategy.

And then leveraging AI to really scale that and produce content at scale in order to really move the needle on awareness and SEO and those sort of things.

Social media, so lots of opportunity there.

And then the other part which marketing teams tend to often forget is really the demand generation portion of it, where I think there’s a huge opportunity for marketing folks to partner up with sales organizations and really leverage AI for prospecting, and really filling out the top of the funnel which marketing teams are tasked with.

You mentioned content at scale.

What does that look and feel like in real life?

Yeah.

I mean, to me, I think if you look at any typical marketing organizations where they’re tasked with producing content, what typically they end up doing is producing depending on the size of the organization.

It’s the written word, blogs, articles or videos, that sort of thing which will come out.

Volumetrically, it’s directly proportional to the size of your team.

But I think with AI, I’m seeing more and more that that is not really tied to the size of your team anymore.

I think the, in fact, one of my clients had four communications professionals as part of the team.

One quit, one was let go, one was in a maternity leave that left one person and they were freaking out about how they’re going to meet their content goals.

Well, AI came to the rescue.

So once you use the right tools and technology and provide the right level of input, the output that one person was able to produce and review was able to cover some of the gaps in resources that they had as a result of team changes.

So we’ve talked about the concrete use cases.

How about cool companies in the space?

What are either incumbents or upstarts bring into the game that’s got you excited?

Hard to pick a few.

How much time do we have?

Go as long as you want.

Yeah, no, I’m kidding.

I think there’s so many, to be honest, and like I said, at a very foundational level, if you’re not using ChatGPD or Gemini or Perplexity, any of these LLMs just for your day-to-day, I think you’re missing out.

So I think I would start there.

One level above from that, which is often kind of overlooked, is just the way to automate routine tasks that you do.

So, for example, ChatGPD, you can create custom GPTs in the paid version of ChatGPD, and you don’t need any coding knowledge.

It’s sort of interesting.

You use GPT to create those custom GPTs.

And so any sort of repetitive information that you typically do, you can really make huge gains out of that.

So I host a podcast for Spark Novus.

The podcast is called the Marketing AI Spark Cast.

So it’s just the routine process of coming up with kind of a podcast brief for the guests, and post-recording, looking at transcripts, and coming up with campaign materials and that sort of stuff has been dramatically sort of accelerated as a result of just using that custom GPT.

So some things at that personal level.

And then when it comes to marketing technologies, there’s all kinds of tools, Descript for video or Jasper for content creation, on brand and on voice, brand voice content creation, where you can sort of upload guidelines and knowledge and really train the AI for that.

And one of my favorite tools that I’ve been lately playing with is this tool called goalcast.io, and it’s a B2B event platform.

But they launched a very interesting product called Content Lab, and what it does is it allows you to sort of remix, quote unquote remix content where you’ve already created a video, whether it’s a video or a webinar, and then it really allows you to stick in the URL or upload the video, and it allows you to create campaign materials based on that pre-existing video content, like mini clips for social or social media written content or emails and blogs, simply by providing that content.

So sometimes a lot of marketing teams go out looking for new content with nothing wrong with that, but there is already so much rich content within the teams that can really be, quote unquote, remixed and repurposed using some of these tools is pretty incredible.

What does the tech stack look like five years from now?

Where’s all this headed?

Current course in Speed, I think he used the image of a hairball, a bowl of spaghetti.

There’s so much going on.

I almost feel like we need a position in the organization just to keep track of the tech stack and teach people how to find what.

Maybe that could be an AI agent.

I don’t know, but where does this all end up?

Yeah, I think two thoughts there.

One is, I think there is certainly going to be an era of consolidation where in the tools, the net native world, there’s lots of products which are really, if you really zoom out and have a look at some of those things, they’re really features of a larger platform.

I think it’s just a matter of time when this AI bubble pops or portions of it pop, where there’s going to be this consolidation.

It could be some of these smaller companies that are extremely capable and built out that capability, getting sucked into more substantial platforms.

We’re already seeing some of that stuff, because nobody wants, you don’t want eight separate platforms doing the same thing.

So from a workflow standpoint, you don’t want to do something in one platform, take the output of that, stick it manually into another platform, import it in, and so on and so forth.

Very tedious.

So I think the way I’m sort of picturing it is in those platforms which have more of a end-to-end workflow, whatever that may be based on the role and the function, those are the platforms that may end up acquiring smaller technologies where there is capabilities that they don’t have, and it’s the classic build versus buy.

So instead of them going building it out, they’ll just sort of acquire the smaller firm.

So there’s going to be some tech consolidation there.

And then the second aspect I see is, I think you said it, agents, where it’s already happening where you’re going to end up with the ability to format the form factor, right?

Based on one set of actions.

So for example, if I’m going to run a campaign about something and I want to write a brief, who is it for?

What industry?

What’s the ICP, you know, firm or graphics, what have you?

What’s the purpose of the campaign?

What’s the key message?

Some key creative, you know, that sort of stuff.

And I want that campaign in, you know, multiple touch points.

I want it on, you know, all the social channels, you know, but the way I posted on LinkedIn may be different than the way I would, I may post it on TikTok.

TikTok is one of my channels.

And then, you know, long form contents, if I want to do a blog or if I want to do a landing page, if I want to do a webinar.

Today, that workflow is quite manual and tedious.

There are technologies, like I mentioned, that can sort of ease some of that, like Goldcast or Jasper, for example.

But honestly, I see that some of these technologies is going to have agents where you can put the brief, set it, forget it, and then out pops the various assets, which would require, you know, human oversight in terms of reviews and those sort of things, of course.

But the ability to create that is going to be kind of dramatically improved and accelerated.

So that, to me, I feel agents is going to be, you know, a big piece of the equation.

And then 100% right, you do need, you do need someone to, you know, keep track of all that stuff.

And, you know, the AI world, it’s moving so fast.

There’s an influx of information.

It’s really hard to, there is no, you know, I’m in this world and there are time people ask me, Hey, Aby, have you heard about, you know, whatever technology, blah, blah, blah.

And the answer is no, it’s just not humanly possible to sort of keep up with it.

So I wouldn’t be surprised that there’s going to be a time where people are not going to care what is out there till the time the problem presents itself.

So very often when I’m talking to CMOs and I talk to 100, you know, hundreds of CMOs in order to help them adopt AI, the adoption of AI for what part is often forgotten, right?

Like where people are like, Hey, I want to adopt AI.

And I’m like, okay, for what?

To do what?

You know, that part is often forgotten.

So I think we’re going to get to a point where, you know, people are going to have pieces of technology, marketing technology to sort of help them meet the standard needs of a marketing function or function for sales and demand, brand and everything that a marketing department does.

But outside of that, there’s going to be a need to where people are not going to bother about what’s out there.

I think the pain comes first and then, you know, they’ll go looking because, like I said, impossible for people to go find, you know, the world of AI is getting, you know, vast and deep by the day.

You’re painting a very clear and believable picture of what the future looks like for the technology.

Can we talk for a second about the humans?

Let’s do it.

Rather than just the blunt, like, do jobs go away?

Is this an automation or augmentation play?

How about, how do you see skill requirements changing or jobs changing?

That strikes me as the more interesting question.

Yeah.

I mean, in a lot of ways, I feel back to fundamentals.

Like it’s almost like the most important skill for people is going to be the ability to learn and relearn, right?

That’s so important because, you know, a lot of times people’s insecurities and we’ve heard stories about jobs going away and that sort of stuff.

It’s all legitimate, but I think it’s just people getting used to the idea, having an open mind, they’re just used to the idea of doing things differently.

In my mind, there is no doubt there’s going to be, you know, a change in the way jobs are structured.

We are going to lose jobs.

We’re going to gain jobs.

But making sure that, you know, marketers and professionals really have an open mind as to the role of AI, how it’s trending, saying tuned in and sort of educating themselves.

So that to me is super important.

And there are so many resources out there to do that.

So to me, I think my first advice to people would be have an open mind in terms of, you know, learning.

Don’t be overwhelmed.

Just take first the first baby step towards educating yourself about what this is.

And then the more you end up knowing, the less daunting and threatening and, you know, anxiety-ridden your life is going to be.

I could continue this conversation forever, but I know you have places to go and people to see.

So I’ll put a bookmark in there and hope that we can speak again.

But thank you so much.

It’s been a pleasure to spend some time with you and we appreciate your insights.

No, my pleasure.

This was a lot of fun.

Absolutely.

Let’s do it again.

Thanks as always for listening and watching.

Don’t forget to give us a five-star review on your podcast player of choice.

And we’d really love it if you’d also take a few minutes to leave a review.

At the end of every episode, we like to ask one of the LLMs to get their take on the episode.

So hey there, chat GBT, OG back in the house, what’s happening?

This episode, we’re talking about the most important things to know about how to use AI effectively.

So what do you think?

Business leaders need to focus on understanding I’s potential to streamline operations, improve decision-making and enhance customer experiences.

They should prioritize ethical considerations, data quality and continuous learning to stay ahead in the rapidly evolving AI landscape.

Now you’re in the know.

Thanks as always for listening.

We’ll see you next week with more AI applications, discussions and experts.