Unlocking the Future of Data Management and AI with Muhi Majzoub at OpenText World 2024

This is Techron tv. Hey folks, we're back at OpenText World in Las Vegas, and we're here with my friend Muhi, and we're gonna talk about what the tech and the platform and how it all comes together. And I have a question for you that I've been thinking about ever since I watched the keynote and a couple of presentations and I can't make up my mind.

Okay. Are we kind of just managing data the same way we always managed it at scale, at a different level of scale? Or is the way that we manage data about the fundamentally change and, and do we need to kind of rethink the whole way we approach data management?

Yeah, Managing data is still the same as we've always done it. The innovations we've added on it is advancing how we capture metadata, which is critical and important, and how we bring gen AI to the data to extract intelligence automation and allow the agents to do more automated work than the human are doing in the past. And that metadata is critical because a lot of the data we have is unstructured.

So the only way to apply any kind of sense of something that feels semi-structured is through the metadata. Do I have that correct? That's correct.

The metadata is Critical, Especially when we operate in a multi-cloud environment. When we are exposing a document that has come out of SAP from a receivable or a payable, we have the metadata sitting at the source with that document, so we're able to gain all the intelligence from that content and that metadata. So how does that manifest itself in the titanium X platform?

You guys have been working on that now, I think for the last two years, but yeah. Um, what are the innovations and, and, and how do they manifest Titanium X? We have done innovations at multiple level from day one when we announced titanium, even before titanium X was to bring all of our product to the cloud and to advance the features, the capabilities of how customers can use and extract value from our different platform.

We have a, a large portfolio of platforms. Uh, they operate in different areas. We have our modern work, which you could call intelligent content, you could call, um, document management, which manages the life cycle of information.

We have supply chain, we call it business fabric. You could call it EDI, you could call it, uh, value added networks, uh, all of that. Uh, we have advanced every one of these platform with new features and new capabilities.

And now with titanium X, we're bringing more SaaS capabilities to even make our SaaS applications more powerful and at parity with our private cloud. But we are also bringing gen ai and every one of the demos you saw at OpenText world yesterday in our CEO and CTO's keynote, and today in my keynote and Shannon, every one of these demos highlighted new capabilities that gen ai or new features that we brought to this platform, The agents themselves. Um, if I look at the stack and if everything we're gonna interface through the entire stack through the agents, will the, the silos start to collapse a little bit?

Because I'm, from my perspective, I'm just gonna be invoking an AI agent and I may not care what's on the back end of that. I'm just trying to get the workflow done. So is everything gonna kind of converge into a single platform or how does that look as we go forward?

So we're, we either converge into a common platform or we do deep integration and you saw, uh, and you will see in our demo area examples of both. Let me start with converging in a different platform and give you a real product example. Our core document, uh, digital asset management is built on the exact same platform as core content, core case, core signature, core capture in digital asset management in core, you can actually click on one of the menu option and now you're looking at a business workspace in core content and seeing how your content is captured, how the meta metadata is extracted.

So that's an example of how we, uh, bring products onto the same platform. They leverage similar agents, they leverage similar restful services in areas where we do deep integration. You saw an example of that in my keynote this morning with service management when Steven Ludlow spoke to how he's bringing content and documents from core content today, those are two different platform, but they're deeply integrated.

So to the user, a common user interface and a common set of standard for restful services or for an agent model allow us to deliver that capability. Mm-Hmm. And how will the evolution of the AI agents occur?

Because it seems to me there's a lot of them and they'll need to be orchestrated at some point to create some sort of workflow. Is that something the human does or is there some software that sits on top of that that will provide that capability? Yeah, So in the early, um, for agents to make decisions, a hundred percent of the time is still early.

It is the technology is not yet there. We need it to mature a little bit more. We need hallucination to be reduced a little bit in some of these gen AI model.

Uh, we are doing two things in one area. We're still letting the human be in control in several area. We're experimenting where you see the agent takes over and do, and you saw an example of that in my keynote.

When Tal Levy, uh, Joseph, our VP of engineering, uh, for the software delivery platform took five manual script and allowed the agent to automate them and create a one automated script of the five manual. That's an early stage work that we are doing. Uh, we are experimenting internally at OpenText on our engineers.

So we are the first user. Shannon Bell, our chief digital officer and CIO highlighted that she is the first user and user number zero. Every time we deliver a new innovation, she deploys it internally either for her team, for our team, or for our professional services and support organization.

We experiment, we ensure the accuracy of the responses, we ensure the performance, we fix any defects we discover in the process, and then we start taking it to beta customers. And after beta customers, we take it to general availability and make it available to the 120,000 customers we have in our portfolio. When I talk to some customers, they're, they get the concept, but they're struggling with how to figure out how to insert these AI agents into their workflow because they are probabilistic.

And they'll say that they may not get the same answer every time, and somebody has to validate the answer and their workflow is deterministic. It has to be the done the same way over and over again 100% of the time. How do we meld those things together?

Yeah. Uh, and, and your comments and statements are very true. You can ask the same prompt of a model and today get two different answers.

Uh, it's getting a lot better than it was last year when Chad GPT first got announced. It has improved hallucination have been reduced, but have not been completely, uh, resolved and gone. Uh, so we are still experimenting.

Uh, today. That's one of the reason we also experiment with four, uh, large language model. We experiment with OpenAI slash copilot.

Mark highlighted that in his demo on stage and gave an example. We experiment with Gemini in our core content aviator. We experiment with LAMA and we experiment with mist trial, and we're evaluating three areas in each of those experimentation.

We look at hallucination slash accuracy of the responses. We look at the cost of deployment and the return on investment to customers, and we look at performance, uh, and we are trying to identify the best set of standard that we then bring to all of our aviators. I also think I heard that you're building some small language models that are domain specific for different use cases.

Is that something you're thinking about? We are not building, uh, language models internally at OpenText through our partnerships with Microsoft, with Google, uh, and leveraging some of the open source models that are available. Lama Misre, uh, those are the one we're experimenting.

We are not building any new language models at this time. Hmm. And your keynote, you talked about the roadmap for titanium X.

Um, so what's coming next? What should we look forward to in 2025? Um, I'll group them into three different areas.

AI and the Aviator is one, every one of our aviator would advance in functionality. We are currently working with so many customers who have either implemented and are up in production or are doing proof of concept through our Earn Your Wing program with our professional services. You, you saw an example of that in my keynote by, uh, Ms.

Sheila Wu. Um, we are taking all that feedback and advancing the AI and the aviator and including incorporating some of the customer feedback, fixing a defect if in if something is discovered or adding a new feature or a new K cap capability. Uh, another example of that, at one point in Mark's keynote yesterday, you saw the the person demoing selecting which workspace to include or exclude.

That's a feature that came from a customer and we are very proud of these partnerships. We have the second area you're gonna see us embed more and more security in our product. Security is stop of mind, and because we use 60 of our product, they better be secure.

We are supporting a total of 300,000 plus customers in sm uh, in our SMB and in our enterprise group. We better deliver the best security. So security will be my second.

Category three will be SaaS. Uh, we're very excited based on customer demand in calendar 2025 in the July, October timeframe, you will see us deliver core communication. We are the leader in that market today in the customer communication, and often customer are using our product and don't even know they're using an open text solution because they're getting it through a partner.

I'll give you an example. If you are a member or a customer of Chase banking, if you have a credit card or if you have an account, your statement is delivered through the open tax platform. Now you say Muhi, how do you know that?

I know that because I work for OpenText and I know they are one of our customers, but the average consumer in the United States in any of our 50 states does not know that's an OpenText platform. We do and we're proud of these partnership. If you get notification from airlines or from your, uh, Comcast saying there's service interruption, you might not know behind the scene is the, uh, core messaging platform in the private cloud or the public cloud because one of these customers is one of our customer and have deployed our services.

You will see us deliver a lot more of SaaS application in our demo area. We're highlighting and showcasing our AppWorks Aviator. AppWorks is our platform known for low-code, no-code today with Gen ai, you could give app work and Excel spreadsheet and and app work.

The gen AI aviator will read the Excel spreadsheet and develop an application for you that without writing a single line of code, it, uh, I've had three customers in the hallway stop me to tell me how great that feature is in AppWorks. So we're very excited and you will see us advance every one of our platform that I showed on screen in my keynote, uh, the wheel, and we will advance aviator security SaaS application and more feature richness. One of the things I think I saw was that there are 15 aviators right now and probably more to come, but there are nearly a hundred agents and more to come.

Yes. What Is the difference between what an aviator is and what an agent is An aviator at time? That's a great question.

Thank you. Um, an aviator can use an integrate into multiple agent. So if you are bringing, and we saw an example of that I believe in Mark's keynote where you saw data coming from Guidewire.

You saw data coming from SAP, you saw documents coming from core content or extended ECM, um, this is the aviator, integrating and assembling content from multiple sources, the concept of multi-cloud and taking that and bringing it to the aviator to process and then present the data back responding to the prompt given. Now you're also, as I understand it, revamping the user interface because the way that we interact with agents and the data's changing. So what went into that thinking behind that?

So several things. Another great question. Thank you.

We're very excited about our jato ui. 2, which for us is April of calendar, 2025. Uh, Jato will bring a new user interface to all of our public cloud application.

The goal of jato was to make our user interface embed aviator in it and make it very visible where Aviator becomes the first touch point of that product, depending on the task you wanna do. Uh, its document management is service management is software delivery. Uh, I am in the soc in the security operations center and I wanna look at XDR, um, whatever the capability is.

The goal of jato is to expose the aviator and make it very easy. That's one aviator. Uh, Jato also has another goal.

We are improving a lot on our accessibility to where we target more and more of our user base. We want everybody and everyone to be able to use our product. Three, we're embedding notification in the product where the product is now communicating certain things back to the user of tasks that are important.

They could be reminders, they could be messages, uh, they could be integration into other application like, uh, open text messaging, core messaging or core MS SMS. Uh, we want to improve in the functionality to where the product or the dashboard you enter into can accommodate as many of your workload in one place, not having you go to five different places to do the same amount of work and many more. All right.

So what is your advice then to customers about how to proceed with this journey? Because a lot of them are, it's it's a big platform and they could be easily intimidated. So is there some place to start that makes sense?

That's another wonderful question. Um, the recommendation I give always and many of the example with the customers you saw sitting either with me on stage or with Todd Cone, our president of Worldwide Sale, or with Mark Beche, our CEO and CTO, every one of these customers started with us coming to say, I have a challenge or I have a business problem. What can OpenText do to help me?

For me, the approach I take with every customer interaction, and I have many in every week, every month, every year, um, that kill deliver the partnership you saw with Lindsay Fe Fernandez, the VP from Catalan. Uh, we started by understanding their challenges and then bringing the best opportunity before they even sign to see what the product can do with their own data. That's building a proof of concept that's very similar to what we do with Earn Your Wing today for our Gen ai, we don't tell the customer to commit to millions of dollar multi-year contract.

We say, give us 30 days, give us a small amount of money that cover the cost of the infrastructure we're gonna deploy and we're gonna give you in the private cloud or in your own data center, the infrastructure needed to deploy Gen AI and to deploy it on top of the platform that you want. And if we don't show you the, the return on investment or today many customer call it return on equity, if we cannot demonstrate that to you in 30 days, then you have zero work to do. We pull out that POC is shut down and we move on.

Um, and so far it's paid dividends for open tax. Customers are excited. Uh, every customer that saw the Content Aviator demo in Mark Beche is coming to see I House that type of information.

It's contract data. Can Aviator do for contract what it did for RFPs in Mark Baron, she's keynote and I say, absolutely yes, because we're bringing the AI to the data. Your data doesn't have to go anywhere.

All right, folks, you heard it here. A, the data doesn't have to move and BROI in 30 days, see if we can match that anywhere else. Hey, movie, thanks being, thank you so much for your time and thank you for being with us this week.

My pleasure. Looking forward to maybe Nashville. We'll see.

Yes, look forward to that. All right. We'll be back in a minute folks.