Swimming With Sharks: Enterprise GenAI Unplugged - S3 Episode 1: Stephen Stouffer

Episode Overview: In this episode of Swimming with Sharks, Kevin Dean, CEO of ManoByte, sits down with Stephen Stouffer, Director of Automation Solutions at Tray.ai. Stephen discusses his journey from marketing automation to becoming a leader in AI integration and shares valuable insights on leveraging AI for enterprise success.

Introduction: Kevin introduces Stephen Stouffer, who brings extensive experience from various industries, including enterprises, agencies, startups, and nonprofits. Stephen has been deeply involved in the AI space, focusing on how organizations can harness AI to streamline operations and gain a competitive edge.

Interview Highlights:

• AI Accessibility: Stephen emphasizes that generative AI is for everyone, from beginners to technical experts, and encourages organizations not to be intimidated by its potential.
• Enterprise Challenges: He discusses the importance of feeding AI with the right data and context to maximize its effectiveness, stressing that while AI can simplify tasks, thoughtful deployment is crucial.
• Tray.ai's Role: Stephen explains how Tray.ai’s rebrand focuses on integrating AI into workflows securely and efficiently, introducing features like the Merlin connectors that enhance data privacy and automation.

Key Takeaways:

• Start Small: Organizations should begin with simple AI implementations, focusing on teams that are open to change.
• Security and Context: Ensuring data security and providing AI with proper context are vital for successful enterprise AI adoption.
• Future of AI: The AI landscape is rapidly evolving, and organizations must adapt quickly to stay ahead, with AI becoming an integral part of daily operations across industries.

Transcript

0:02

Hey there and welcome to Swim with Sharks, a deep dive into generative AI for the enterprise. I'm your host, Kevin Dean, CEO of Mantle Byte. And I am thrilled to have you join us on this exciting and dynamic world of artificial intelligence for the enterprise. So we're excited to have with us today, Steven. Steven is from trey .ai, name change there. Steven, welcome to the show. Why don't you tell us a little bit about yourself? Thank you. Thank you. Yeah. The rebrand was, was, was fun to go through. but yeah, my name is Steven Stofer. I'm the director of automation solutions here at trade .ai previously trade .io. We'll get into that a little bit here, in a second, but, yeah, my, my roots, is marketing automation. And then that, that was like about a decade ago. My, kind of entered into the space under marketing automation, email marketing that kind of blossomed into, revenue operations. And then over the last several years, man, I've just been diving into the AI space. lots of fun things. I've my background is on the enterprise side, I've worked for a company called Dex Media. If you remember the Dex phone books, and then they merged or acquired Yellow Pages. So I've been very much on the enterprise side. And then I've also worked at agencies and startups and nonprofits. So I've kind of seeing the gambit when it comes to organizations. Well, that's so cool. It's nice that you have such a broad experience. So tell me, right now, there's so much conversation about generative AI. What's one surprising thing that you've seen that's going on and something that you think people should know about? Yeah. I'd say it's like, it's for everybody. Like, I think there's some voices out there in the market who talk about all of the crazy, awesome stuff that you can do with, with artificial intelligence from, you know, embedding into your platforms and machine learning and, all, and it can kind of scare people, right? It can be like, that's, that's not for me. I'm not a technical person. I don't know how to work with servers or APIs or anything like that. But, you can like any level of, of, where you are from your journey with an AI from a total newbie who doesn't know how to code, you can get a free open AI chat GPT account and start using the generative AI piece for generating content and ideation and learning about ways that you can incorporate it into your role, all the way up to the engineering type folks. So if you are a super technical person, you can build some really, really cool stuff for your organization. So I guess I would say don't be scared of it. and dive right in, it's the future of organizations, it's the future of the way that we're gonna do work. So don't be scared, it's for everybody, not just the engineering. You know, I absolutely love that. And that's a good way to, you know, kind of frame our conversation. AI is for everybody now. Like it's kind of leveled the playing field to some degree. But then would you think that while it's leveling the playing field, there's still this thing where enterprises can take it to the next level to kind of, you know, say, now I'm going to separate, use it to separate myself from the competition. Yeah, I mean, absolutely. mean, even Trey here, we saw that as an opportunity to separate ourselves from some of our competitors. So previously we were Trey .io. Trey is an IPass, an integration platform, so you can plug in all of the amazing tools that you have in your organization and orchestrate that and get things connected in a really good and complex and custom way. But we also saw an opportunity where like, look, AI is coming. of the marketplace, that is an additional tool that folks are trying to make heads or tail of and kind of get behind and integrate into your systems. So not only are we building our own native tray AI tools where you don't have to get another subscription, you don't have to get authorization from another company and go through compliance. It's just baked right there into the platform. But then you can also plug in, you know, something like OpenAI or Bedrock from AWS. So yeah, it's definitely an opportunity for companies to set themselves apart. And if you don't, you're going to be a step behind the curve. I like what you're saying, right? And so it gives businesses the opportunity to set themselves apart, you know, to get ahead of the game. I think that there's this interesting concept. To and I'd love to your take on this, you know, you mentioned that now gen AI is for everybody I was literally talking to a friend who used it to help their kid, you know schools back in session They use it to help their kid solve some math problems, right something that they didn't understand easy to help with that So that's one kind of area of it But when it comes to the enterprise, okay now we're talking about something completely different now. We're in the enterprise space Sometimes I think that organizations seem to think that, it's just as easy as helping my kid with their math problem on school. Is that correct? Is it really that easy when you're looking at implementing GNI in the enterprise? Gosh, that's such a loaded question. I'm gonna go with it depends. So it depends in that if you're trying to empower your teams to do simple tasks like math problems or maybe coding would be a good example where you want to enable your developers to develop JavaScript, HTML, CSS, Python, whatever languages that you are. Sure, they can go in, they can do prompt if they know exactly what they want. it in. But where it gets difficult, and this is the maybe piece of it is, is AI is only as good as your data and it's only as good as the context it has. So do you want your marketing team to be saying, know, chat GPT, create me an email for this product, but it can't create the email unless it knows what is your corporate tone of voice? What is your products and services? Who are you targeting? What's your target audience? emails have performed in the past really, really well, and can we use that information to dictate the email that's being created? yeah, mean, out of the box, sure, you can have a very rough version of generative content being created, but if you really want to unleash the power of it, you have to feed it data, you have to feed it context. And that's where kind of the magic happens with AI in deploying it to maybe non -technical folks who don't know that they need to feed it this context. So that's on the enterprise deployment team to be setting those guardrails around security and feeding it the context that's appropriate for the teams that are leveraging it. That's very insightful, right? So what I'm hearing is that, you know, while it can make your life easier, there's some work that has to go into it in order to get it to the place where now you can start, you know, leveraging it across the enterprise. I think of it more of like a scale of usefulness. if you don't do any pre -work and you don't do anything, yes, you can use it, but it's not going to be as helpful if you do the pre -work and you fix the underlining data, if there's data issues and you can feed it the context of what is your corporate tone of voice and, how do you word things and how do you position your products and services and, know, what would a, an upsell look like in your customer buying journey? so. So it just becomes more useful the more time and effort you put into it. But that's the great thing about AI is it is useful out of the gate. It just might not be as useful if you were to have a little bit more of a thoughtful approach to deploying it within your organization. Yeah, I love that. You really need to have a thoughtful approach to how you're going to deploy it in your organization in order for you to get the biggest bang for your buck. That makes a lot of sense and that's pretty smart. What other challenges do you see when you're talking to organizations about taking the next steps with GEN .AI and how they can use it to automate their business processes? What other challenges, you mentioned data is a challenge, but what other challenges do you see organizations running into? our routine link. Yeah, I mean, I think... The main challenge is it can do so much. It, can be difficult to whittle that down to what the most impactful thing it can do for your organization. You're to have a lot of different teams that do a lot of different things. they operate differently. So figuring out where you start is as oftentimes a challenge in that it can be actually so difficult that people just don't even start. Like they don't even do anything because they're like, gosh, this is going to be way more difficult than what we can handle. So. I guess my advice there is like start simple, start with a beta test group, maybe start with a specific team that maybe is conducive to change more than other departments, right? So like your legal department, your HR department, your maybe corporate IT department, they might be a little bit slower to adapt than maybe your marketing team, right? Your marketing team around content generation, they might be more quick to adapt to testing it out. But also security, right? So you do need a loop in your IT team. You need to go through the checks and balances. If you're SOC 2 compliant organization, you need to make sure that you're adhering to all of the controls that you've agreed upon for your audits. But yeah, would say like starting small because that can be very intimidating. And then also we talked a little bit about data, right? So data and fixing the data is it can be a big blocker to launching something like AI in an organization. it's just not clean. Yeah, that makes a lot of sense. That's very practical. Now, for our audience, one of the things that we want to do is help them understand the tools and the platforms that are out there. Can you explain to our audience what exactly does Tray .ai do? Yeah, so Trey is an integration platform. It's an Ipass solution, so integration platform. And then it connects your tools. So just to give you a very simple use case, let's say you have Salesforce or HubSpot or whatever your CRM choice is. And let's say if you have a HubSpot, a deal goes closed one, or if you have Salesforce, an opportunity goes closed one, and you want to do something with that information. Let's say you want to, one, give a little shout out to your team. You want to send like a Slack notification or a Teams notification to be like, hey, so and so closed the deal. It was like this. This is a really good opportunity for the team. Get some corporate communications out. But then also maybe what you want to do is you also want to take the opportunity and impact the customer experience, right? So you want to send out an invoice. So you need to create invoices. You need to share that information with QuickBooks or NetSpeed or whatever your ERP is. And then you need to communicate. financing you need to make sure that there's follow -up. Make sure they pay their bill on time, right? So, Trey is kind of the platform behind the scenes that orchestrates the communication between all of your tools. And that's just one use case on maybe a quote to cash type, use case, but there's like tons out there from marketing, having an ingestion for leads, right? So you might use, Google or Facebook or LinkedIn lead gen forms for lead capture. but getting that connected into your tech stack easily is really difficult. Normally you have to go into LinkedIn, you have to log in, you have to export a CSV, you have to import that CSV into your CRM. It can be very time consuming. Trey basically handles the automation from start to finish. So you can plug LinkedIn into Trey, plug your CRM into Trey, and when someone fills out a form, it's milliseconds of that record getting into wherever it needs to go. Well, that's pretty cool. Now, you did mention early on that Trey's gone through a rebrand focusing on AI. Can you tell us about some of the features that Trey now has from a generative AI perspective and how that might work within one of these workflows or for an organization? Sure, yeah, well, So we've released a series of connectors called Merlin connectors. So for an example, we have Merlin guardian, which tokenizes your data. So if you're looking to do something with, let's say, open AI, let's say you want to, write an email for a specific person, you know, you have their name, you have their title, you have their, address information and you deploy something to marketing for them to use. They might actually put all. that information into their prompt and say, Hey, write an email for, and then they put in sensitive information, PII sensitive information. Merlin guardian, essentially, if you were to put the build this on the trade platform, you can push it through Merlin guardian. It'll actually tokenize that information. if it sees, marketing team put in a name, they put in a last name, they put in an email address. We do not want to send that to open AI. So Merlin will essentially tokenize it, send the tokenized version of their prompts to open AI. open AI will have the context of what those tokens are, right? It'll know it's a name, but not know what name it is. It can write that email, come back with the response and tray. And then Merlin guardian has another feature where it de tokenizes it. So it converts it back to that personalization. And then it sends that de tokenized version back to your marketing team. So they don't actually know that their sensitive information, they fed into the platform, never actually made it to open AI. but the experience is seamless, right? It doesn't throw an error message. It doesn't say, don't give sensitive information. It is all handled on the backend securely, open AI, never got access to it, but you can deploy it in a way where you've set those security guardrails for marketing. That is an amazing piece of functionality, especially now in this world where we have to be very cognizant about security and privacy and data, what we're sharing and how we're sharing it. That's an awesome feature and functionality that can be built into your workflows by using Trey, right? So when you think about... just when you look out into the space in general and you see what generative AI can do and there's so many things that it can do. What are some features or capabilities that you believe are missing in some of the tools in the enterprise space? Yeah, I would say security might be a concern. I think that folks are moving very fast. And that's the reason why we've been built the example I told you to begin with is because folks are moving so quick. It's difficult to trust that some of these smaller AI shops who might do the thing you need it to really, really well. But they're very small team, right? They may not be sock to compliant. They might not be HIPAA compliant. They might not be GDPR compliant. So you know, making sure that you've built those controls on your end versus depending them to is big. And then the other big one that I would say is, you know, early on in this conversation, we talked about like a knowledge base or context around prompt building. You know, unless you've built your own chatbot on OpenAI, there is an opportunity to upload knowledge base articles, but they don't make it very easy to add that additional context. So I think that that could be bit of a gap right now, but everyone is scrambling to fix that so Yeah, I'd say that like security and then also the context piece of prompt generation Yeah, those are two great things that need to be addressed, but I think exactly like you said, like those are things that people are scrambling to fix as quickly as they can. So what about the future? What do you think some cool things are gonna come in the near future? What excites you about, you know, what's next? Yeah. It's interesting question. It's hard to predict what the future is going to be, but we are definitely at the, exponential growth AI journey, right? I mean, it, when did open AI release chat GPT like late 2022. And here we are only a couple of years later and almost every company, every conference is talking about AI. just within a couple of years, it's completely reshaped the landscape. So another two years, I don't know what that's going to look but I know that like you mentioned earlier, it's level setting the playing field. So the way that we do work is going to be different. So if you're a developer, your, your day -to -day workload is going to look different. It's going to be the co -pilot of every employee in every organization. And you know, here in Dallas, we actually have McDonald's that's fully autonomous. There is no person who hands you your food. There actually is no person you can even talk to when you go into the McDonald's. So if you want to look at a weird AI future, take a look at the McDonald's in Fort Worth because it's completely autonomous. Now that's an extreme example. I don't think every restaurant in America or the world is going to be handled by robots in the near future. It's definitely going to be ingrained and there's going to be basically HR onboarding courses for, you know, there's gonna be like risk mitigation and there's gonna be like your AI certifications. So it's, going to get interesting real quick. That is very cool to think about going to McDonald's and not seeing a person at all. And honestly, some of the McDonald's I've been through, I'm excited. Like, can you please get back to McDonald's next to my house? You're like, I want my burger and french fries as quickly as possible. I don't care if it's a robot that gives it to me. Yeah. Yeah. man, this has been a fun conversation. Hey, you know what? Can you tell us a little bit about what you do when you're not geeking out over Gen .ai? What type of fun things you like to do? man. All right. So I'm I love rockhounding and I don't know if you know what that is, but basically it's what you say. No, me about it. want to know. I don't know. So rock counting essentially is you basically looking for like special like certain minerals like gold, silver, pyrite, like quartz, like, you know, you kind of like for like fun rocks and gemstones kind of out in nature. So that's like that's really fun. I'm like big into nature. that is very, very cool. Steven, thanks for joining us on the show today and thanks everybody for tuning in to Swimming with Sharks. And please check us out past, post, future episodes at manobyte.com forward slash SWS for Swimming with Sharks. Thanks everybody and we'll see you the next time.