This podcast is brought to you by Fast Slow Motion, a team of expert consultants that have helped hundreds of growing businesses build scalable systems on the HubSpot and Salesforce platforms. To learn more, visit fastslowmotion.com.

Amy has, like you’ve said, it’s been the buzz. And so our team we’ve traveled to Chicago, San Francisco, Boston, Atlanta, and it feels like everywhere in between. And we’re having one conversation. And that conversation is business leaders asking us that same question — What is AI and how can I leverage it effectively in my business?

Welcome to the Fast Slow Motion podcast, where we teach you how to build a business you love while enjoying your life. In our podcast, we talk key learnings, cheat codes, hacks and principles that have served us on our journey and the hopes that they serve you as well. I’m your host, Sam Donaghey. And today I’m joined by Chad Seales, Fast Slow Motion’s Chief Technology Officer.

Hey, Chad, how are you?

Good morning. I’m doing great, Sam. How are you?

Awesome. Very, very well. Thank you so much, Chad. In the first few episodes of the podcasts, we’ve been sharing principles that can help business leaders build a business that they love while also creating an amazing culture. While the ratings today, I’m happy to say that we’re doing something a little bit different. We’ll be talking about AI and we’ll also be talking specifically about how AI tools within the Salesforce and HubSpot CRM platforms can help business leaders really create a business advantage — create an edge — and also how they can get started with that. Chad, there’s just a ton of buzz and excitement about AI. I suppose people can get really lost in the noise. Can we start with the question — what even is AI?

Yes. That’s a great question. It’s really been a whirlwind since November 2022 when ChatGPT was released. AI has, like you’ve said, it’s been the buzz. And so our team, we’ve traveled to Chicago, San Francisco, Boston, Atlanta, and it feels like everywhere in between. And we’re having one conversation. And that conversation has business leaders asking us that same question — What is AI and how can I leverage it effectively in my business?

So when we have the conversation, the best place to start is really to begin with the what. So what is artificial intelligence? Because it’s actually far more than just ChatGPT. It’s been around a long time and it really lives in three forms. There’s machine learning, deep learning, and then generative AI.

So machine learning is a system that learns from data, improving as it learns more, as it gets more data, as it has more response to it. Deep learning is a form of machine learning. But whereas machine learning might use thousands to millions of pieces of data, deep learning could use millions to billions of pieces. So it’s just more complex and it’s leveraging neural networks to do this. And where you might see deep learning is image processing or speech recognition — so it’s a more complex reasoning.

And the last form, the one that’s buzzy that has everyone talking, is generative AI, which — generative AI is just that — it’s generating new things. It’s the content creation. It’s in conversation, able to produce something that’s never existed before.

So each of these forms — machine learning, deep learning, and generative AI — they have their own strengths and they have their own weaknesses. And it’s important to note that AI is not one tool. It’s really a toolbox. And a business leader needs to understand when they should reach inside of that toolbox and pull out which form of AI to leverage for their various needs.

So to make it practical — because that’s a lot of theory, a lot of definition — let’s make it practical. We have a client that’s a hardware and software B2B company, and they leverage all the different forms. Machine learning — they’re using it to score their opportunities. So how likely is an opportunity to close into a new customer? Taking a set of data, leveraged on that data, you’re able to make predictions on whether that opportunity is going to ultimately turn into a customer.

Deep learning — they can take a photo of the hardware that they install at facilities and then they can identify: is there an issue with this hardware that we’ve seen before? Surfaced from their own proprietary captures of hardware malfunctions. And then ultimately the goal there would be to surface — this is the issue, here’s how to resolve it.

And then generative AI — their salespeople are leveraging their CRM. Inside of their CRM, they are leveraging generative AI to produce new personalized sales emails. So one company across all the different forms of AI, kind of dipping into the toolbox, getting real value from it.

Wow. Okay. I love that. So it’s more of a toolbox. Thank you for giving us a premise on what’s in the different types of AI. Are there any emerging trends that you find particularly interesting?

Oh man, that’s fun, Sam. Yes. The past year has really uncovered some of the mystery around the trends. There’s just so much noise, so much buzz, so much excitement. And that was around the potential of the technology, but also the risk of the technology. People were leveraging it, but they didn’t understand it. And so they could see that it was materially different and that there’s a ton of opportunity, but they were unsure of the how.

And so what I’ll tell you, Sam, is just trend-wise — in a day-to-day business person’s life — things aren’t materially different than they were a year ago. People are still doing generally the same things. AI has had an impact, but it hasn’t been the seismic shift that we expected. It just hasn’t changed that quickly. That’s not to say that it’s not coming. And that’s not to say that the impact that AI is currently having isn’t immensely useful and effective. It’s just that it’s taking longer to really catch hold in the marketplace and for businesses to truly begin leveraging it well.

So you asked about the trend that I’m most interested in. Really today, the way that it’s taking hold is that people are leveraging AI for task-based work. So a task like a prescriptive task — maybe how likely is this opportunity to close? Which opportunities should I focus on? Tell me which one and then I’ll go work that opportunity. Or a generative task — like I need to create some marketing collateral, so create a blog post for me, create the outline, create the social caption. So you can see that it’s task-based.

And even in the non-generative and predictive — with things like bots. So bots are AI, but they’re still very task-based. So you might want to cancel your hotel reservation — a bot can handle that request. And a bot can even be a little bit more complex, a little bit more intelligent. Let’s say you wanted to interact with a company about their software and you wanted to demo their software — well, a bot could interact with you, you could put in your information, the bot will capture your interest and then could even feed back to you that demo. But it’s still task-based.

And so currently, the trend that we’ve seen is AI is solving for task, and it’s solving for task really well. So this is exciting. There’s a lot of opportunity for business leaders to leverage this, to be able to do more and to do it excellently.

But the trend that I’m excited about isn’t this task-based phase that we’re in. It’s this new phase that’s coming that’s not task-based, but really grander. It’s objective-based or goal-based. Technologists are calling this agentic AI. So agentic AI is technology that’s profoundly different — while still the same — because here’s what it’s able to do. Agentic AI can take not only the task, but it can take the why and it can form what tasks need to be accomplished to actually reach this objective.

So let me make it practical. Currently the way people are using AI is — I, as a marketer, want to create content. So I go in and I use AI to create the content. Well, soon a marketer will be interacting with an agent — agentic AI — and say, I would like to become an influencer in this space. Now craft the campaign and all of the appropriate actions that I need to be taking to achieve that influencer status. And the agent would be intelligent enough to understand the dependencies upon the tasks and to state — these are tasks that are best for the human, and these are the tasks that are best for the agent — and then go and actually work in tandem with the human on that initiative.

So that’s the trend that I’m most excited about. It’s taking the task-based functionality that we have today and marrying it with something exceedingly more powerful that’s really going to drive business in a new direction.

Chad, that’s so, so interesting. So there’s this agentic trend that you’re talking about — while we’re not quite there yet, what’s kind of withholding us from getting to that point now? Is AI still learning behind the scenes until it can get to that stage?

Yeah, that’s a great question. It’s just the technology. AI is a budding technology. It’s existed — forms of AI have existed for a long time, over a decade. We’ve been interacting with forms of AI. But that specific functionality is still developing, it’s still growing. We can see it, we can begin to interact with some forms of it. But it doesn’t have the intelligence, it doesn’t have the power that it needs yet.

But what we’ve seen is the cycle time for AI technology is very rapid. And so that means — if you follow OpenAI, you’ve seen GPT-3, GPT-3.5, GPT-4 — it’s quickly iterating on itself and becoming more and more powerful. So it’s just a matter of time and the sophistication of the technology.

So that’s a really interesting trend that business owners are going to want to pay attention to. What would you say the most common concerns are when we start talking about AI? Because rightfully so, people can get a little bit worried about artificial intelligence and where it might go in the future.

There really are a few concerns. The first of which is hallucinations. So sometimes AI can generate misleading or incorrect responses and data. It gets a little bit too — we call it creative. And it’s funny — generative AI is a lot like a really good salesperson. They’re extremely confident but loose on the facts. A real-world example I saw — I read an article in CNET that they were having to issue a bunch of corrections on a couple of articles that they had written the outline of with GPT, and they had to go back and write corrections because a substantial amount of the article just was fallacy — it wasn’t true. And so a generative AI tool can lie to you. And so it’s important to understand that a generative AI can hallucinate, and ensure that you have a human in the loop who’s fact-checking the technology.

And then the second concern, other than just hallucination, is really data security. With AI — whether it be machine learning or deep learning or generative — the question is: what is this technology doing with my data? Who has access? If we’re talking about machine learning, it’s who has access to the data that I’m putting into the system, who’s consuming it, and then what purposes do they have outside of just giving me back a prediction? And when you talk about generative, it’s everything that I feed into that conversation — what level of access do they have to the conversation? Are they storing it? Are they storing it and using it to train their systems?

So if I’m having a conversation with generative AI about a proprietary process that we have at FSM to help our clients — well, is the technology going to store that prompt and that conversation and then learn on that for the next person? And so it might have our proprietary processes or our proprietary data inside of their model. So understanding the data security implications with AI is really important.

And just to make it practical — here at FSM, we’ve laid out a really simple policy for our team. We take data security and privacy very seriously. And so we have told our employees that no one at Fast Slow Motion can use their FSM email accounts to log in to all of these GPT systems. Additionally, employees can’t supply or upload customer information that’s not publicly available into any of those tools. And we provide to our employees — the ones that make sense — we provide them access to the tools that we have vetted from their request. So our policy is: don’t take our data, don’t take our IP and put it out into a system that we don’t trust. Let us vet and ensure that we trust it as an organization, and then we put it behind a paid subscription to ensure that our employees aren’t compromising themselves or our customers or our company.

Okay, great. So it’s crucial that you take steps to safeguard your business when utilizing these tools. Chad, you mentioned earlier about CRM platforms and how they’re incorporating AI into some of their platforms as well today. And I think that’s really going to give businesses incredible advantages in the marketplace. Can you give us some insights on how CRM platforms like Salesforce and HubSpot have been working behind the scenes to bring AI tools to their customers?

Yeah, it’s really interesting, Sam, because they’ve taken really different approaches — materially different approaches. HubSpot took the approach of getting as much functionality, as many features, as much value to their end users as quickly as possible. The CEO of HubSpot has this quote where she says that research translates into application in just a matter of weeks. So that means they’re coming up with ideas, they’re learning about new technology, they’re developing the techniques to actually leverage that technology, and they’re putting it into the hands of their users in a matter of weeks. It’s a very short cycle.

And then on the other spectrum, you have Salesforce. Salesforce’s approach is ensuring that the AI that a business is leveraging is trusted and secure. So what they’re wanting to ensure is that a business is able to get the benefits of AI without inherently increasing their risk. So HubSpot — as quick as possible into the hands of the end user. Salesforce — trusted, secure, a little bit slower. But at the heart of it is the security of the data and the platform and the vulnerability of the business.

Chad, you’ve shared so much great information already and I can’t wait to dig into the rest of this topic with you. For our listeners, thanks for tuning in to part one of our discussion on AI. Be sure to tune in for the rest of the conversation in Part Two. Remember to subscribe to the podcast on your favorite podcast platform, or you can find archived episodes at fastslowmotion.com/podcast.