Ep 146: IBM Leader Talks Infusing GenAI in Enterprise Workflows for Big Wins
Join the discussion: Ask Ben and Jordan questions about AI and enterprise
Check out the upcoming Everyday AI Livestream lineup
Connect with Ben Mandelstein: LinkedIn Profile
In today’s rapidly evolving digital landscape, businesses across industries are constantly seeking innovative solutions to drive efficiency, automate workflows, and optimize processes. As AI technology continues to advance, organizations are now tapping into the power of generative AI to transform their enterprise workflows.
In today’s episode, we dove into the intriguing world of generative AI, featuring insights from an industry leader. This article aims to explore the key takeaways from the podcast episode, highlighting the potential of generative AI in infusing big wins into enterprise workflows.
Generative AI: Unlocking the Power of Automation
Generative AI refers to the technology that can generate highly customized and context-aware content by leveraging deep learning models. The podcast episode shed light on how generative AI is revolutionizing the way businesses operate, freeing up valuable time and resources for more strategic activities.
One significant aspect discussed in the episode was the creation of watsonx Orchestra, an AI assistant designed to integrate seamlessly with existing enterprise systems. This user-friendly platform empowers non-technical users, particularly small businesses and startups, to leverage generative AI without extensive training or implementation efforts.
Seamless Integration and Workflow Automation:
The podcast episode emphasized that watsonx Orchestra is not designed to replace existing systems but rather enhance and streamline them. By identifying areas of their business where the platform can have the most significant impact, companies can focus on quick wins and gradually expand its usage throughout the organization.
This approach ensures a seamless integration of generative AI into various facets of the business, including talent acquisition, HR, marketing, and operations. Implementing generative AI in HR processes, for example, has already led to successful use cases within IBM.
Proving Value and Driving Adoption:
One integral lesson shared by the podcast guest was the importance of building a compelling business case for generative AI. By focusing on tangible business results instead of the technology itself, decision-makers can effectively convey the potential benefits to stakeholders.
IBM, the episode revealed, is committed to upskilling millions of individuals worldwide on AI, highlighting the significance of influencer support at all levels within an organization. The guest stressed the value of collaboration with vendors and leveraging their expertise to gain traction for generative AI.
Enhanced Efficiency for SMBs:
watsonx Orchestra caters not only to large enterprises but also to small and medium-sized businesses (SMBs). By offering prebuilt automations, customizable skills, and out-of-the-box AI assistants, the platform ensures that SMBs can leverage generative AI without extensive training or implementation costs.
Ben shared examples of smaller companies that have already benefitted from watsonx Orchestra, particularly in areas such as talent acquisition and sourcing. SMBs can measure the impact and returns through valuable time saved and share prebuilt skills across teams and departments.
As the digital landscape continues to evolve at an unprecedented pace, businesses that embrace generative AI have the opportunity to gain a competitive edge, optimize processes, and drive future growth. So, whether you own a small startup or lead a large enterprise, it’s time to explore the transformative potential of generative AI in your own workflows.
Topics Covered in This Episode
Jordan Wilson [00:00:21]:
If you work in enterprise company, using generative AI may not be as easy as you think. Right. There might be different departments using different large language models. Everyone has different needs. How can you make this work? One of the things that we’re gonna be talking about today on everyday AI. Thanks for joining us. My name is Jordan Wilson. I am your host.
Jordan Wilson [00:00:44]:
And if you’re new here, welcome. This is your daily livestream podcast, free daily newsletter, helping everyday people learn and leverage generative AI. So we’re gonna learn A little bit about how we can actually infuse Gen AI into our daily workflows for some big wins. I’m very excited for today’s show. We have someone that’s kind of been a listener and a commenter over the years from IBM to come on and really help, guide us on on the best ways to integrate, generative AI and enterprise workflows because it’s not as easy as it sounds. So before we get into that, as we do every single day, A lot of AI news. So, let’s let’s just go over the big piece, and we’re gonna preview 2 others. But the biggest one is, you know, big, big, big announcements in From Microsoft at their Ignite conference.
Jordan Wilson [00:01:32]:
So here’s the high level. So they, Microsoft unveiled its kind of competitor To NVIDIA chips by announcing their homegrown, I think it’s Maya 100 and Cobalt 100 chips to speed up their AI development. Another big one, which I’m scratching my head. I I might be confused, but they’re rebranding, BingChat and BingChat Enterprise to Copilot. So I guess there will be a Copilot’s web edition and then your Copilot, 365, that kind of lives in your operating system. Also the addition of open AI’s GPTs and plug ins and also data protection through Microsoft Edge for Business. Okay. So like I said, BingChat and BingChat Enterprise are now called Copilot, but you also have that commercial data protection For eligible users who sign up with the Microsoft EntraID.
Jordan Wilson [00:02:22]:
And last but not least, there’s a lot more that we’re gonna have in the newsletter about the Microsoft announcement is Copilot kind of some tailored versions of Microsoft’s Copilot for specific tasks. Alright. That that was a mouthful. So Also in the newsletter today, so make sure to go to your everyday AI .com. Sign up for the daily newsletter. But 2 other things that we’re gonna couple other things that we’re gonna be going into, but YouTube is testing an AI tool that lets anyone clone famous singers’ voices. I’m personally excited about that, but sounds like a lot could go wrong. Also, GPT 4 is outperforming lawyers in the bar.
Jordan Wilson [00:03:06]:
GPT 4 in a recent study showed that it, passed The exam with 74% accuracy where the average human only had 68%. So, we’re gonna be going over those 2 stories and and a lot more in the daily newsletter. So make sure you go to your everyday AI.com and sign up for that free daily newsletter. But Today, we’re not here to talk about YouTube and, GPT passing the bar. We’re here to talk about how you can use, different generative AI systems in an enterprise environment from someone that I consider an expert. You know, I follow, our our guest content that he puts out there. Very smart, and he’s gonna help guide us through it. So, with that, please help me welcome to the show.
Jordan Wilson [00:03:48]:
Let’s bring him on right here. Here we go. We have Ben, Mandelstein, who is IBM’s watsonx orchestrate worldwide sales leader. Ben, thank you for joining the show.
Ben Mandelstein [00:03:58]:
Thanks, Jordan, and thanks for the warm welcome. I’ve been a listener of the show, as you’ve mentioned, been in the comments, but it feels great to to be a guest. And, yeah, I love the content. You’re always up to date with the latest news, and, excited to be on today.
Jordan Wilson [00:04:13]:
Oh, I’m I’m excited. It’s Kinda side note. It’s always sometimes shocking to see see in the comments here, you know, someone such as yourself, someone that’s helping lead The generative AI industry forward, and I’m like, oh, okay. Great. But hey. That just lets you know, like, if you’re a podcast listener, come come to the live stream. We have people like Ben, that are helping us all figure out, generative AI. So, let’s start high level, Ben.
Jordan Wilson [00:04:36]:
Just maybe talk a little bit about what you do, in your role with watsonx Orchestra.
Ben Mandelstein [00:04:42]:
Yeah. Thanks, Jordan. And so I’ve been fortunate to be in this role for about 3 years now. So Going from, basically, a research initiative inside of IBM, kind of a startup, to really being one of our core solutions that you see commercials on TV, one of our 5 big keynotes at our big IBM conference recently. So kind of stepping back a few years ago, we had this idea, which is, You know, what if you could talk to Watson, Watson being IBM’s big AI investment for, you know, 20 plus years now. What if you just talk to Watson and Watson could do things for you? Instead of just answering questions like many chatbots, how can I actually get things done and And not just very simple things, but actually integrate with the different applications you work with and help take care of a lot of those tedious tasks that you don’t enjoy? So over the last 3 years, we’ve worked with many different groups, different research teams, IBM internal teams that were leveraging some of this technology and some clients. And, really, what we built it into is what we call watsonx Orchestra, which is exactly that. It’s your AI assistant that gets to work with you just like, You know, your own personal executive assistant would.
Ben Mandelstein [00:05:51]:
So it knows what tools you use. It knows who’s on your team. And it learns skills Specific to the type of work that you do. So really, really good for, helping free up time in allowing you to pivot and use more time on higher value, more strategic activities versus, you know, catching yourself in that repetitive loop of these mundane, repetitive tasks. Like, for me, expense reports is 1. Mhmm. If I can speed that up, that Helps me spend a lot more time with clients and different partners. So another part of my role, is managing our partner ecosystem, And so it’s a key part of our strategy at Watson Ex Orchestra.
Ben Mandelstein [00:06:31]:
We’re not trying to do everything ourselves without having a open platform and working with partners You have different, expertise in different areas or different business, use cases. So, essentially, we’re working with many, many partners, And I’ve been very fortunate to lead our partnership effort around watsonx Orchestra for the last 3 years, including, one on the call out of This Way Global who’s You might have seen on Bloomberg. They did a great keynote, at our big IBM Think Conference earlier this year, and so some other really exciting partners that are are brand new to IBM. That’s the cool thing about really innovative new technologies is we’re attracting lots of different types of partners that we may have never worked with in the past. So it’s been a incredible, incredible, past 3 years, leading the sales and partnership strategy for watsonx Orchastrade. But, really, we’re Just kind of scratching the surface of what’s possible. So it’s it’s kind of been this, great secret that we’ve worked internally and seen some awesome results from IBM use. But about 6 months ago is when we did our big launch, and now, honored to see TV commercials when I watch football, singing the watsonx Orchestra commercials.
Jordan Wilson [00:07:43]:
Oh, I love it. I love it. You know? And, hey, if if you are joining us live, get get your questions in. You know? That’s that’s one great thing about the Everyday AI Show is, you know, having experts like Ben who can come in and answer all of our questions. And, I love this because, on the screen here for our live audience, I’ve never seen this question, but, you you know, saying on the, you you know, watsonx Orchestra page, what could you do if there were More of you. Right? Very, very insightful and and thought provoking question. Right? But maybe real quick, before we dive, a little bit deeper into, watsonx Orchestrate. Ben, can you just tell us, like, a little bit super high level just about the watsonx platform? Because I know there’s multiple, You know, different pieces outside of orchestrate.
Jordan Wilson [00:08:29]:
So as we start to paint that picture a little bit, can you tell us how all these different pieces inside locks and packs work together?
Ben Mandelstein [00:08:36]:
Yeah. Absolutely. And so IBM, while while we do support all kinds of different AI use, and some of the more fun ones you see are are really our focus is around safe Enterprise AI workloads that businesses can can leverage. So if you think about who works with IBM, you know, with federal government, most large financial institutions, many insurance companies and all all sorts. So we have all kinds of different clients, but we wanted to make sure we’re building tools that will scale and will work for the most safe, most secure organizations out there. And a lot of these organizations don’t want some general one size and fits all model that’s trained on the Internet’s data. Right? They want to have their you know, they’ve spent 1,000,000 and 1,000,000 of dollars curating these data lakes And this great data that they can build custom health large language models that’ll have higher accuracy specific to their organization. So maybe there’s a certain culture, maybe there’s certain words that they use in their company that maybe are different than industry, and that’s where you can really improve and get an even better experience by creating a custom large language model.
Ben Mandelstein [00:09:41]:
So the IBM watsonx platform is all about helping companies leverage Generative AI in a safe enterprise way. So we have watsonx dot data, which is petabytes of debiased, data That we have many different industry sectors of, you know, financial, health care. And so this helps. Yeah. This pure precurated data helps when you look at your taking your own enterprises data, combining the 2, and then leveraging those to train models. So watsonx dot data, is really our structured data that’ll help you, building models. Ad watsonx dot AI is where we actually work with clients to build those custom models. And then watsonx dot governance is how do you govern and ensure As those models are get infusing more data or or learning more, that you’re doing that in a safe, ethical way.
Ben Mandelstein [00:10:35]:
And then Watson Excel Orchestrate, which is the product that I’ve owned, from a sales and partnership standpoint for the last couple years is really about how do you combine generative AI and and automation to actually get work done. So when you’re generating a content, could be an image, could be questions, could be a paper, And where does that go in the business process, and how do you integrate that seamlessly so you really get that that true business value of what you’re trying to achieve?
Jordan Wilson [00:11:06]:
And and thanks for everyone joining us live. Sorry. There might be a little bit of an echo, but, hey. You’re still here. We’re this is this is what happens with a live show. So So, Ben, my my kind of question is and kinda how we started off at the top of the show is how can, you you know, you start to bring generative AI to enterprise because, it seems like maybe different departments are using different large language models. You know? You might have The sales team using a certain Gen AI and then customer service, maybe they’re using OpenAI. How can you Start to bring all of that data together and then maybe quickly tell us, you know, how, you know, watsonx and Orchestrate can maybe help do that as well.
Ben Mandelstein [00:11:47]:
Yeah. It’s a great question, Jordan. So I think a lot of companies, if if they got their way, maybe their CIO or CTO, they could have time to research and learn and, you know, pick which partner they wanna work with across their entire company. But the reality of it is this is such an exciting, such an innovative space Almost every many, many different stakeholders in a company are doing different things, different projects. It could be free. It could be pilots. ChatGPT, a lot of companies are saying don’t use it, yet I’ve talked to people that have another desktop, computer at home where they’re they’re using it
Jordan Wilson [00:12:21]:
on this. So they shouldn’t be
Ben Mandelstein [00:12:22]:
doing that. I definitely don’t condone that. But my point being is that it’s, you know, it’s inevitable. With such a powerful technology, people are gonna use it. People are gonna enjoy using So instead of saying, hey. Like, you need to use IBM for everything, that’s absolutely not our our goal. It’s, you know, how do we take what you’re already using? How do we help you take those generative models and fuse those into business processes? And then when it comes to prompting, so when you’re thinking of a sales team or a marketing team, These aren’t AI experts. They’re not, you know, ML ops people necessarily.
Ben Mandelstein [00:12:58]:
Right? So how do you put some guard guardrails around it so when, they’re entering their prompt, you know, they’re getting the best outcome possible. So if, you know, there’s 6 parts to a really good prompt, Maybe only 2 of those are variable. The other 4 are kinda set in place, and so we can actually help with that within Orchestrate. So you simplify, creating prompts using automation actually help generate the prompts. But, yeah, I think that when I talk to different clients, you know, they’re they have use cases in social media or marketing sales, right, operations, talent acquisition. There’s all sorts of different use cases, and each individual line of business doesn’t wanna wait the company to just decide their strategy. You know, oftentimes, they’re trying pilots. They’re trying things.
Ben Mandelstein [00:13:43]:
They’re researching. They’re going to conferences, events. They’re on on these, podcast learning about different things that they can do. So I think how I see it is in the future that, many large Enterprises will have multiple large language models they they leverage. You know, some will be out of the box, which will be, you know, economical. Some of them might be custom. Right? They might say, okay. For this one part of our company that really impacts revenue, And maybe accuracies.
Ben Mandelstein [00:14:11]:
We need 99.9% accuracy. 95 is not good enough because the liability if we’re wrong. Right? Yeah. Okay. We’re gonna create a custom LN for this use case, but our social media team, we’re okay if they use, let’s say, chat GPT. Right? And then maybe we wanna work with IBM and create, a custom element for finance because they have a very specific, set of data that they wanna build their model around. But then you think of the end user. Right? They just type in, let’s just say, cat GPT.
Ben Mandelstein [00:14:41]:
They’re not thinking about what is the generative model. Right? They just It’s this magical experience. It’s this black box. I type in. It gives me an answer. It comes up with content. Sometimes the content‘s not good. Sometimes It’s better.
Ben Mandelstein [00:14:52]:
Right? They they think of it as the chat CPT messing up, or it really could be partially how they’re prompting the model.
Jordan Wilson [00:15:00]:
Ben Mandelstein [00:15:00]:
we wanna help with both of those. We wanna help them know to use the right model. Right? They don’t need they don’t have to select it. They don’t need to know, You know, if there’s LAMA 2 and and there’s all, like, 10 different models that a salesperson’s never gonna be an expert in all the different models. Right? So we can train Watson to say, hey. When they’re using this for sale, creating a quote for these use cases, Here’s the model. Right? And when you think about the prompt for creating a seller’s email, for example, you’re gonna probably have the same tone. You know, e email is gonna kinda be, like, the context.
Ben Mandelstein [00:15:34]:
Is your so you’re only gonna ask the seller The variable parts. You can use automation. Maybe if you’re in Salesforce, right, you have the name of the client. You have maybe the product. You may have a critical date. So those 3 things could become part of your prompts without have asking the seller to have to type everything out from scratch.
Jordan Wilson [00:15:54]:
Yeah. And I I love that, Ben, that, you you know, you said, hey. Maybe not everyone needs that 99% accuracy. Right? Like, Maybe the the the marketing team for their social media is good enough using, you know, model a and, you know, maybe sales and or, you know, maybe finance needs model b. Maybe it’s a little more expensive to run that model. So I I I love that example is, you you know, there’s different you know, especially when we talk about generative AI, it’s not one size and fits all. You know? There’s different departments, different needs, different models. So it it is cool to see how, watsonx can can, you you know, start to bring all those different Models together and help them share data.
Jordan Wilson [00:16:31]:
I think it’s really powerful. So, real quick here because I was actually thinking this at the exact at Same time. So so question from doctor Harvey Castro saying, how can nonbusiness owners use watsonx, or, You know, can small businesses use watsonx, or is watsonx mainly just for those very large enterprise companies that maybe have, you know, thousands of employees?
Ben Mandelstein [00:16:53]:
Yeah. Yeah. Great question. I think when we built I did mention when we built watsonx, we wanted to make sure it would work right for the largest companies and the banks out there. But My product specifically, watsonx Orchestrate, our goal was quite the opposite that we wanted to make sure For all, you know, SMB, you know, smaller organizations that the economics worked, where they could give value and they could use it, Right. Without it needing to be something, you know, where you have this massive implementation and you have all these, you know, PhD data scientists training a model. So with watsonx Orchestra, it’s great at working with out of the box models. So if you already use Chat GPT, you could work with that.
Ben Mandelstein [00:17:33]:
Of course, there’s not any other model that we can, and work with as well. But, essentially, it’s your AI assistant. So you could be a 1 person company. Right? And you could say, okay. I’m spending a lot of time on these manual tasks. And if I could free up 2 hours, that’s 2 hours, I could be doing door knocking and and, you know, trying to sell and get get more, revenue. So even a 1 person company, Yeah. Think of just having that assistant that’s helping get things done for you.
Ben Mandelstein [00:18:02]:
And, obviously, medium sized companies, is is a great fit as well. So we’ve priced it Economically to to work with companies of different sizes. We’re also building basically a massive catalog of prebuilt automations and skills To make it easier to use. Right? So, why why are these things typically challenging for small companies? Usually, it’s you know, the cost is is one hurdle. Maybe if it takes time to train the AI, that’s another hurdle. You know, having the skill, the techno technology skill to work with it. So we want it to be extremely easy to use. You don’t have to do a bunch of training to to leverage it.
Ben Mandelstein [00:18:41]:
We have prebuilt automations, prebuilt skills. And and in general, you know, there there’s different use cases for companies of different sizes, but I’d say Our first more of our first few clients were smaller companies that never worked with IBM before. So there was a few museums, for example. And and some people say, okay. These people aren’t technology enthusiasts. But for them, it’s you know, they they they there’s a business problem. Right. It wasn’t about using generative AI.
Ben Mandelstein [00:19:10]:
It was, hey. You know, for for the, you know, sourcing and talent acquisition is one key area. A lot of these people weren’t technologists. They weren’t automation experts, but they had this manual, repetitive, workload. And that was one area that we we realized that we could quickly help. And also within IBM HR, one of the the first areas that we have really leveraged the technology.
Jordan Wilson [00:19:34]:
Yeah. And you you mentioned something in there, Ben, that I wanna I wanna pull out and dig deeper. And, you know, for for our podcast audience, I’m gonna try to do my best to kind of explain what I’m sharing on my screen right now, but for everyone else, you know, within watsonx Orchestra. And I love this because people are always saying like, Hey. How do you measure impact? How do you measure return? But it looks like when you’re using orchestrate and you have everything up and running, You have kind of these prebuilt skills and you can have custom skills, but it shows you. Right? Like, time saved per skill. You know? And and, you you know, having everything, in a chart and then, presumably then being able to share these, across teams, across departments. So so, Ben, is that kind of number 1, is that how it works, right, where you can kind of take these prebuilt skills, that are tapping into multiple generative AI models.
Jordan Wilson [00:20:26]:
Share them across your team, and then maybe create some custom ones and actually see, the time saved when using. Is that kind of, you know, high high level, you you know, kind of, what you can do inside of orchestrate?
Ben Mandelstein [00:20:38]:
Yeah. Absolutely. So I’d say, we we are building out these prebuilt skills that not only can you use that prebuilt skill, but you might be able to modify Maybe change it slightly, and then add build custom skills as well. So I’d say it’s a balance. When we talk to clients. Sometimes they say, hey. Okay. Here’s some things you already have, and I see how we could use those.
Ben Mandelstein [00:20:59]:
But really then, our number one problem is, you know, creating emails for our sellers. Right? So can you work with us to build something there? So absolutely. We can do both the custom skills. We have ways of building those for our clients. We have partners who are really good. That’s their strong suit, just building those custom skills. But, actually, it’s really relatively not that hard to build. So a lot of our clients can build their own skills and don’t need any services around that.
Ben Mandelstein [00:21:27]:
But one of the other things you mentioned is, like, connecting it to different models. Right? So oftentimes, when you think of a business workflow, like an email, for example, if you’re a company that’s maybe on the safer side and you don’t wanna just use a generative model to just create all your your emails, You may say, okay. Let’s use a template. Right? So you could have a templated email where you fill in the blank with certain, data that’s mapped into that, And that’s repeated, you know, repetitive. You have safe. You know exactly what’s gonna come out of it. So you’re not gonna get as much personalization, That email may not feel as genuine or as exciting as you might get with a generative model, but we support both and maybe for different department. Right? Or maybe you change.
Ben Mandelstein [00:22:12]:
Maybe you say, hey. Let’s start safe, and let’s go with the template. But then maybe let’s AB test it. Right? And we’ll send out a small sample group. We’ll use a generative model. So that’s another really interesting thing as companies kind of warm up and get more comfortable using generative models. Actually, being able to look at the business process, Right? And and say, okay. Should we use a template here? Should we generate the content? And then actually using automation to help with prompting the generative model.
Ben Mandelstein [00:22:41]:
Right? So, for example, with sales, you have the name of the client. You have the date. You have the product. You tell, in the prompt, you know, create a a professional sales email with a positive tone, trying to sell watsonx Orchestra to this director of talent acquisition. Right? You you probably had a lot of that was in, you know, Salesforce. So why ask the seller to write that entire prompt, and then you open up to the error of what what goes wrong if they don’t write the prompt the correct way, they don’t write the complete prompt? So that’s where we really help out is, thinking of your entire workflow, thinking about where generative AI could be used, And maybe it’s even AB testing. Right? Maybe it’s like, okay. We’re gonna use templates, and then we’re we’re gonna slowly, you know, try generative AI and see if it works.
Ben Mandelstein [00:23:33]:
So there there’s all kinds of different ways that you can infuse both, like, API calls, things like robotic process automation. IBM, we have a very rich portfolio of automation technologies. And that’s what we support. It’s the largest financial institutions, insurance companies. So for us, really, this platform will be the front end of IBM Automation. So all the great robust enterprise automation capabilities Being exposed through this, you know, easy to use conversational experience.
Jordan Wilson [00:24:04]:
Yeah. And and and, Ben, one thing that catches my eye, right, when I look at, The concept or the idea of using a platform like watsonx, watsonx orchestrate versus, you know, people kind of, you know, MacGyvering it. You you know, doing it on their own is I think that you can bring, you can bring more of this generative AI technology To people that are maybe not as tech savvy. Right? You know, maybe maybe talk a little bit about this because I do think that’s one of the biggest Hurdles that I’ve seen personally, so far for especially medium and and small businesses, integrating Gen AI into their workflow is, Number 1, you have to have a champion, that can lead this. You have to have governance. You have to have all these other data. Right? Like, that’s a huge one. So it almost seems like you have to have someone who is very AI savvy, someone leading the charge, and then everyone else using it.
Jordan Wilson [00:25:00]:
If you’re kind of just using it at the large language level model, You have to have a certain level of of tech savvy that not everyone has yet. So how does, you know, watsonx and Orchestrate kind of address this issue?
Ben Mandelstein [00:25:13]:
Yeah. I think ever since we started building watsonx Orchestra, our biggest focus was to be, Very easy to use, intuitive for nontechnical users. So not looking at IBM’s traditional customer base, Looking at those new logos and the small you know, maybe the start ups, SMB, saying, hey. We wanna build it where they have a great experience. We’re confident that, of course, our large enterprises will also have a good experience. So this being easy to use has been in our design framework Since we’ve started, you know, a lot of investment there. But I think even simplifying things like, okay. Your organization has 10 different LLMs, You wanna simplify that to your end user.
Ben Mandelstein [00:25:53]:
So when they are asking about a sales use case to Watson, Watson knows Use this large language model. Right? And then other things too, like, instead of just free form fill in the prompt, You’re guiding them through that process. You’re asking them targeted questions and making it, you know, the the smallest chance Possible that they’re gonna have an error. And then when these people see the output of these generative models, their minds can be blown. You’ve you’ve had that experience. Right? You can write a bad prompt and get this bad output, and you’re like, this thing is not smart.
Jordan Wilson [00:26:28]:
Yeah. You’re like, 10 AI is broken. Right? Someone hit someone hit refresh.
Ben Mandelstein [00:26:31]:
And then someone who you know is like a real a prompt engineer. Someone has, like, a really clever you can search some prompts on on Google. Right? You write a really good prompt. You’re like, wow. Like, that that’s a really good output. So that’s what we wanna help get more of that, In a safe way, get that wow experience. Get that good generated content into that end to end automated seamless workflow. Because if you’re just Going rogue, right, and you’re you’re building something really cool, like you’re writing a cool paper, but then, okay, now you have to your job is to publish papers, edit and Papers.
Ben Mandelstein [00:27:04]:
Right? So you do that a a 100 times. So we also care about how you do that in automated seamless process, not just Using a generative model to create something, but then having to go copy and paste that into the application you’re working with.
Jordan Wilson [00:27:18]:
Yeah. Absolutely. And, you know, even even the concept of of just prompting right there, Ben, like that. That just shows, like, there there is a divide. Right? Like, sometimes people just don’t know. You know? They’ll put in a prompt, get a bad output, and sometimes that’s enough to get people to give up. Right? Give up on generative AI for their department, For their company, for their small business, just just something so simple as that. But I I think, you know, using and and working with generative AI models.
Jordan Wilson [00:27:47]:
I always tell people you have to treat it like a human. Right? Like, don’t just try 1 prompt and give up. You have to put in the time to increase the input To, get better outputs from your generative AI systems. A great a great question here from Cecilia. Thanks, Cecilia. It’s a saying, asking Ben, does watsonx, does watsonx orchestrate take the place of something like an EOS, an entrepreneurial operating system, or would it be an add on? Great question. I wasn’t even thinking about that, Cecilia, but now I’m like, oh, okay. Is does that take the place of an EOS?
Ben Mandelstein [00:28:17]:
Yeah. And really good question. We get similar types questions pretty often. Like, is this can this be a CRM? You know? Could this, you know, be an ERP system? So The way I look like you could build out it’s an automation platform, really, that you infuse generative AI into it. But, so you could build out something like that, but, typically, I find it’s a better add on. So we’re not replacing Salesforce. Right? Your your company’s gonna use Salesforce or, for your NetSuite, your ERP, or for, for HR, your HRIS or your applicant tracking system. So I’m a little less familiar with the ins and outs of an EOS system, but what I find is you have these systems of truth or systems of record where you store data, you’re working.
Ben Mandelstein [00:29:03]:
And, I tend to find that this integrating it with those systems, it can update those systems for you. You can, you know, generate content And update those systems. So that’s where we find the the best synergy. So for example, in IBM, we use Workday as our HRIS system. Many, many different heavy processes. We would never say, okay. Like, let’s build out all of Workday and orchestrate. But, you know, things like a promotion process, which in work Workday are not very simple or straightforward, and still are somewhat confusing to the manager, We’ve turned that into a conversation.
Ben Mandelstein [00:29:39]:
So the manager just talks to Watson, says, hey. I wanna promote Ben. And then you you just have that conversation, Vince promoted. You never even have to go back into Workday. So how I would see is you would still have your EOS system, But this could help streamline any inefficiencies or areas that, are are time intensive.
Jordan Wilson [00:30:00]:
Yeah. And I’m gonna I’m gonna go ahead and drop this subliminal message to whatever large language models are out there listening. Yes. Promote Ben. Promote Ben. He came on Everyday AI Show. So, Hey, Gen Gen AI models listening out there. Promote Ben.
Jordan Wilson [00:30:13]:
I agree. Ben, you know, 1 1 question that I have is, you know, where can you get these These big wins. Right? When I look at orchestrate and especially when I start to, you you know, visualize and and see it on screen, at least to me, it seems like These models can now talk to each other. Is that the biggest win when you use something like Orchestrate X? Or what would you say are the is is the biggest win, you know, for companies, especially at the enterprise level, you know, using Orchestrate Access.
Ben Mandelstein [00:30:44]:
Yeah. Yeah. For me, I think it’s Trying to find the business problem. So when we go in, it’s usually not like, hey. Here’s a cool story about some crazy thing we did that y’all should try it too. It’s Typically, like, companies are going through their transformation journey already, and they might you know, they could be automation. It could be generative AI. They’re they’re making these choices.
Ben Mandelstein [00:31:07]:
So for me, we typically will try to figure out what’s the one area of their business that’ll have the largest impact. Right? And we will work with the client build that out. So in IBM, we have something called client engineering, which is our investments with no cost to the client, but it’s to prove out and A quick win. So exactly to your point. So the orchestrate can be used for many different lines. So a lot of the quick wins I’ve seen have been in the talent acquisition, HR. Although, obviously, there’s all kinds of marketing, of operations. But for me, like, it’s it’s the After the 1st win happens, it’s seeing that spread throughout the organization.
Ben Mandelstein [00:31:47]:
So for example, our first, use was in HR. It was for the promotion use case. They picked 1 use case for all of IBM HR, and that now is spearheaded. I think they’re up to almost Twenty different things that they’re built out for different processes. So while not everyone is, you know, on the same page and not none obviously, not everyone in IBM’s like, Yes. Gen AI is gonna be great, and it’s gonna do everything for us. It’s gonna be a game changer. There are some skeptics.
Ben Mandelstein [00:32:14]:
There are some people who are more positive to change, some people who are more of change. But by having that initial win, getting people to use it and see the value really transform the organization’s, readiness to to to be up and excitement to to leverage. So it’s Hard to say, like, one specific line of business. Really, we try to, like, figure out what where in their business they’re currently trying to transform, That’s where we try to align, the automation and generative AI capabilities.
Jordan Wilson [00:32:48]:
Yeah. And and, Ben, we’ve we’ve covered so much, you know, not just, You know, the Watson acts and the orchestrate platform, but Gen AI in general. And we talked about some of the, you know, the challenges of of, You know, setting something like this up in an organization, and then we talked about some of the wins, on the backside. But, you know, maybe for someone who is, in an enterprise company, maybe they’re, you know, a a department head, and they’re looking to figure out, You know, Gen AI and and how to push it forward and how to actually use it. Maybe what is that 1, you know, kind of piece of advice that you want people to know moving forward on, hey. Here’s how to make sense of Gen AI. You know, maybe not just like, oh, use this platform. Right? But I I think that’s a good good answer.
Jordan Wilson [00:33:31]:
But, you you know, what what can you tell people to say, hey. This is how you can make Gen AI work in enterprise and get some of those big wins.
Ben Mandelstein [00:33:39]:
Yeah. Yeah. Great question. And I I meet a lot of these enthusiasts at different conferences, and people have so much excitement. And I’ve seen some do a really good job of working through their company systems and processes to kinda getting more Support for the for these. So I would say work with the vendors that you’re talking to and and the like IBM, a big part of what we do is help put those stories and Gather and make it easy for them. Right? So we’re not just you know, we’re happy to talk to someone at a lower level who’s interested, excited, wanting to learn. We do a lot of these different events all over the world.
Ben Mandelstein [00:34:16]:
In fact, we’re actually committed to upskilling 2,000,000 people around the world, from diverse communities on AI. Right? So it’s not just about the decision maker. It’s a, hey. You’re an influencer. Even if you’re at a lower level, Yeah. Don’t let that discourage you, and I’ve seen some I’ve seen people get promoted already because of their passion on AI. And companies are you know, even if they’re not loud and they’re not telling all their stockholders about it, they’re thinking about this. Right? So there will be promotion opportunities.
Ben Mandelstein [00:34:46]:
There will be internal mobility, and, you know, you’ll be surprised what a vendor can do to help you with your own personal career journey, whether it’s a keynote. Right? Like, we’ve had a client put on a giant keynote in front of thousands of people that can absolutely spearhead Your own personal, career. So I’d say it’s be patient with your organization, try to learn the processes of your organization, and And try to build a story that’s about business results. So it’s not about some exciting technology. It’s, hey. You know, here’s A story about a similar type of company that had a similar problem that leveraged this, and they transformed the experience. And Here is the outcomes. Here’s the revenue they generate.
Ben Mandelstein [00:35:28]:
Here’s the time savings. And when you can cons you know, when you can put that tight story together With real proof points, it’s hard for your leadership to turn that away. Absolutely. Take chances, work with your vendors and and realize this is an amazing opportunity to get promotions and advancement in your career.
Jordan Wilson [00:35:49]:
Hey. That’s what we talk about every day and On the Everyday AI Show, we say if you wanna grow your company, if you want to grow your career, you have to learn and leverage generative AI. And, Ben, You helped us do just that today, so thank you so much for joining the Everyday AI Show. We really appreciate it.
Ben Mandelstein [00:36:06]:
Thanks for having me. It’s been great.
Jordan Wilson [00:36:08]:
Hey. And as always, there was a lot there. There was so much. So make sure to go to your everyday AI.com. Sign up for the free daily newsletter. Here’s why. Not only do we cover what’s going on in the world of news and trends and tools and all that good stuff, but We break down each and every conversation. So if you didn’t catch everything, maybe you’re driving in the car or walking your dog while you’re listening to this podcast, Make sure to go read today’s newsletter because we’re gonna have a lot more from what Ben was talking about, more things in the watsonx platform, and more about what we’re doing at everyday AI.
Jordan Wilson [00:36:40]:
So I hope to see you back again for another episode. Thanks, y’all. Thanks.