Scott Rayburn: Hello everyone, and welcome to today's webinar on how to unleash enterprise cost optimization with emerging technologies. I'm Scott Rayburn, and I'll be your host for this insightful journey into all things AI, ML, RPA, automation, and process optimization. We have a great show planned for you today. But before we get into that, let me introduce our topic. According to research from Everest Group, which we'll explore in more depth soon, advanced automation, big data analytics, and generative AI are now among the top five enterprise digital capability priorities for 2024. Enterprises can expect to see productivity gains of anywhere from 20% up to 45% in the software development lifecycle thanks to this emerging tech, but many enterprises are facing major roadblocks and gaps to maximize the potential impact of emerging tech for process automation and, in turn, reduce their costs and increase revenue.
Everest Group sees a lack of vision, planning gaps, governance, data management challenges, and other blockers coming to the fore. But adoption at scale is possible with the right investments, both internally and externally, and working with strategic partners like DataArt. Today's agenda will cover top enterprise concerns for 2024, the optimization tech ecosystem and proven benefits, and the roadblocks and best practices to overcome them. Finally, we'll end with real-world examples of generative AI's impact on the enterprise.
Let's introduce our speakers. First, we have Mayank Maria, Vice President at Everest Group. Throughout the presentation, he'll share exclusive research on our topic from Everest. How are you today, Mayank?
Mayank Maria: Doing great, Scott.
Scott Rayburn: Perfect, thank you. And now I'd like to introduce Yuri Gubin, our Chief Innovation Officer here at DataArt, who will be providing commentary on what he's been seeing on the ground with real clients and what kind of technology investments we're making in this space at DataArt, a leading IT services firm. Yuri, how are you doing?
Yuri Gubin: I'm doing good, thank you. This is a very important topic. I'm excited that we have this webinar, and lots of data and insights will come.
Scott Rayburn: Perfect. Thank you, guys. Thanks for being here, and thanks to the audience for joining us today. So, without further ado, let's kick it off. Mayank, do you want to take it away?
Mayank Maria: Absolutely. I'm going to be sharing my screen. Let's jump right into the agenda you set, Scott. Let me touch upon the top enterprise concerns in 2024. This is based on a survey we conduct at the beginning of every year with leading CXOs of typically Fortune 1000 enterprises.
In 2024, we collected responses from nearly 350 enterprise CXOs. We try to gauge a lot of things from them: What are their business priorities? What are the challenges they foresee? How do they expect their tech spend to change in 2024? And so on. I'm bringing one data point from there for you: around the top enterprise challenges in 2024. We see here that cost and margin pressures have emerged as a top challenge across the board. No surprises there. It's really the pessimism that the world walked into 2024 with, carrying all the baggage from 2023 and the kind of slowdown we've been seeing. Enterprises really want to prioritize cost and margins right now in this environment.
Of course, there are other priorities as well. They want to drive growth, stay agile to adapt to customer needs, and deal with geopolitical tensions. However, the pedestal for cost and margin has really increased coming into 2024. Right now, the flavor is doing more with less, and that's also translating into the technology priorities that enterprises are carrying.
Building on the same survey data, we asked enterprises what technologies they would invest in. Of course, cybersecurity and cloud solutions are at the top, being an enabler for many other technologies and themes as well. But if you see the other three in the top five- advanced automation, cognitive big data analytics, and generative AI- all three have a big role to play when it comes to optimizing processes and realizing efficiencies. Generative AI, specifically, was a new entrant in 2024. In 2023, among the top five, instead of generative AI, we actually had RPA. So, some movement on optimization technologies, but these technologies are very much on the radar.
Quickly on generative AI: the world has really gone big on generative AI investments, and it's heartening to see that most experiments so far have been productive in some sense. Only 8% of respondents said they haven't seen any meaningful impact of generative AI on their businesses yet. A sizable 25% said they've seen improvements in productivity, meaning tasks that were being done manually are now done with greater productivity. So people can work better, more efficiently, and more accurately. Then, 67% of respondents said they're seeing processes getting transformed, meaning human intervention in the process itself is being reduced with the advent of generative AI.
Let's take it forward a little bit and peel this onion further to see what the optimization technologies landscape looks like. We, of course, start with RPA, a technology that's been around for many years. It's mostly effective only around structured data and repetitive tasks, so the complexity it can deal with is limited. Of course, RPA has led to significant gains for enterprises in use cases like data extraction and data update. But the real beauty comes in with the advent of AI and, more recently, generative AI.
You'll notice that we're seeing "plus AI" and "plus generative AI." This really means a lot, because the gains to be realized with AI and generative AI are incremental on top of RPA. Some people mistakenly think that if they adopt AI and generative AI, whatever investments they've made in RPA will go down the drain. That's not true, it's incremental benefits that are to be realized with AI and generative AI. Now, greater optimization is possible around more complex tasks. You could also realize automation around unstructured data, such as data extraction from unstructured documents, more accurate predictions, forecasting, and, more recently, with generative AI, synthetic data generation, information summarization with great accuracy, and more intuitive conversational interfaces. This is a journey. The bottom line is RPA, then augmented by AI and generative AI, is how we should see the tech landscape for optimization.
Now, since we were discussing this gradient and continuum of technologies and new technologies coming in, there are also newer ways in which enterprises are taking initiatives around these optimization technologies. If I look at the bottom layer, here, tools and technologies, we are moving from RPA to now augmentation with AI and generative AI. But there are also changes in other layers. For example, the approach: enterprises are trying to move from a very siloed approach, where any business unit would take their own initiatives, to establishing some processes around centralization, knowledge transfer across business units. That's what enterprises are trying to get to. Also, the primary focus earlier was around low-hanging fruit-whatever we could get our hands on, let's automate today. Now, they're trying to realize some business benefits beyond cost. How can these technologies impact the top line? Can they enhance time to market? Can they elevate the customer experience? In that sense, these are becoming C-suite agenda items, being driven top-down as part of broader digital transformation initiatives.
Yuri, you've been tracking this industry for so long, working with clients on these technologies. What changes do you believe are key in how people view these technologies today compared to maybe three or four years ago?
Yuri Gubin: Absolutely. Not only is it on the agenda right now, but it's also been elevated to the board level because of the potential of AI, Generative AI, and automation to disrupt how things are done and do more with less. It has the potential to be disruptive in terms of efficiency and productivity, letting people work on something exciting, complex, and new, and taking the burden of routines out of their way. That's why it's now elevated all the way to the board level and is part of the broader strategy of company and technology development, and all the transformation initiatives you just mentioned.
I also wanted to say that 2023 was the year of experimentation. Many companies, including our clients, did evaluations and invested in R&D to understand how it works, what the potential is, and whether there is value in it. Now we have so much evidence that there is value, not only in gen AI and conventional AI applications and models, but in combining all three, as you explained earlier. The majority of the customers you talk to – 67% – explicitly said it significantly improves the flow and processes.
This year, instead of just talking about experimentation, all the ideas are being lined up in a pipeline into a roadmap. Every initiative is being structured as a business case. There's a challenge to be addressed with gen AI, and there is evidence everywhere that something from different fields or models can be reused or used directly to solve a particular problem. We talk about how much we do it, we talk about constraints in terms of time, and we always measure the potential value that this will generate. It's not hypothetical anymore, we use examples and models. This is a significant change: it is very prescriptive and focused on ROI.
Mayank Maria: Very true. I think 2023 was about enterprises trying to see how generative AI could work for them, but in 2024, the shift is that they're asking who it has worked for, wanting to replicate and learn from them, and then get on that journey. Hopefully, we'll see many more scale-ups, especially around generative AI, and enterprises realizing significant benefits.
Let's keep going. We discussed a few benefits of these optimization technologies, but if I were to put it into buckets, there are really three:
The first is implementing these technologies to optimize business processes, such as sales and marketing, recruitment, supply chain management, finance, and accounting.
The second is around when companies build software, how these technologies feature there, and what kind of optimization can they offer around tech development and engineering processes.
The third, of course, is that business and tech processes also impact customer experience. In this third layer, we talk about how these technologies can be used to define new product features and extensions that elevate the customer experience.
The third is still pretty nascent. We've seen some use cases of AI, but generative AI-led feature extensions and enhancements are still in their early days. We'll leave that for another day. Today, we'll focus more on the impact these technologies are having on business and tech processes.
Let me jump into the aspect of business processes first. We did a quick analysis, business processes could be around financial accounting, HR, supply chain, and so on. Cumulatively, if you see the impact of these optimization technologies across verticals, there is a clear winner emerging, at least on the RPA and AI front: the BFSI (Banking, Financial Services, and Insurance) vertical. It's really a function of the volume of such processes that these enterprises deal with. The BFSI vertical trumps all others. Other factors include data availability, how much of the process is customer-facing (so there's a greater need to optimize), and the regulatory angle, these technologies are being used to drive regulatory compliance, making sure all processes are by the book. Hence, BFSI is much ahead when it comes to RPA and AI.
For generative AI, the story is slightly different. Enterprises in every vertical are trying to experiment with generative AI today, so there isn't as much of a delta between BFSI and others. It's still early days, pilots and proof of concepts. As we move to production and scale up, we may see clear winners emerge.
I also thought we'd highlight some use cases of these optimization technologies coming together. For example, loan and claim processing in BFSI is a combination of technologies that help elevate the process. Starting with CRM and RPA, which can help with basic data entry and validation, then AI or AI-based chat functionality helps a customer go through the loan or claim application, answering questions and building the application. Generative AI can be used to produce personalized loan or claim documents for a particular customer. In that sense, the journey is completed with multiple touchpoints based on these optimization technologies, each having a different kind of impact.
Yuri, any examples from your customers where these technologies have come alive in processes?
Yuri Gubin: Of course. One quick comment on things not in AI: recommendation engines, production planning, and predictive maintenance. The state of AI models and the industry is such that it's a matter of finding the right model and toolset to get it done. There's very little risk in doing these experiments. Everything depends on data and how mature the organization is, but very high-value projects can be completed with relatively quick and short AI projects. We've seen this in the past, and we are reiterating this year.
Customer experience creates value for organizations and their consumers. It's still being actively measured. In many cases, the technology created with AI and Gen AI is being deployed with a "man in the middle"-not necessarily exposing the customer directly to Gen AI but having customer support personnel use AI tools to quickly process a large number of inquiries in a productive, personalized way. This is happening across industries.
There's also a perception that gen AI is just about generating text or images, but when you work with gen AI projects, you realize there are techniques like vectorization and retrieval-augmented generation. You can use these to build tools and pipelines to automate processes that don't necessarily interact with end users, such as data mining, mapping, and acquisition. This is where we see elements of gen AI being used a lot.
Mayank Maria: Makes sense, Yuri. That takes me back to the classic copilot conversation we tend to have, essentially, AI is assisting humans in doing their jobs better. As I heard you, that's the dominant motion right now, rather than generative AI being the front for the end customer or user.
I'm shifting gears a little. Yuri and I thought we'd also discuss our observations when it comes to using these optimization technologies for the software engineering process itself. Scott, I think earlier you highlighted some numbers about productivity being realized based on these optimization technologies. These are based on yearly returns and data we've gathered from different enterprises and service providers. Anywhere between 20% and 45% productivity gains have been realized in the software engineering process itself. This productivity isn't just about saving costs, it could be about greater code accuracy, faster code development and deployment, and, for software products, improved time to market.
If you look at the different layers – RPA, AI, and generative AI – you see that the value realized by software engineers increases incrementally. With RPA, it's automating code deployment. With generative AI, you get code generation, the classic copilot functionality, helping developers code faster and with higher quality. The biggest impacts are in development and testing, while ideation relies more on human creativity.
Yuri, any thoughts on using these tools with developers?
Yuri Gubin: I'll be brief. We've tried multiple tools to assist our developers. There are different requirements and constraints, and everything depends on the project, language, skill set of your developers, and the problems you're trying to solve. Perhaps a tool will dramatically increase performance, but some developers may try it and not like it, which is fair. As long as there's ambiguity, the sooner you try and see how it works in your project, the better.
Mayank Maria: Makes sense. The context is important, and how you use these technologies in your projects will determine how much productivity you realize.
Let's shift gears again and get into the next chapter of our discussion. These technologies have a lot of potential, but enterprises face many challenges in realizing their full potential. This is based on conversations we've had over time. The main challenges are:
Taking the first step wrong, not prioritizing the right use cases, or lacking a vision for scaling up.
Governance involves measuring the efficacy of a use case, identifying challenges, establishing central practices for tools and technologies, and sharing data across the organization.
Data-centric challenges-data in silos, concerns about data access, ownership, ethical use, regulatory boundaries, and skill set challenges.
Others lack C-suite buy-in, focusing only on cost benefits and not considering other benefits.
Quick recommendations for addressing these:
For use cases: Use a framework to prioritize which processes to implement these technologies. Our framework has two dimensions: improvement potential (contextual to your process maturity and user feedback) and impact potential (how crucial the process is for your business, measured by users, volume, or cost).
Citizen-led discovery: Get suggestions from process users on where they need optimization, rather than a top-down approach.
For governance, the approach scales with your organization. Small organizations may use a siloed approach, but as you grow, move to a centralized approach (COE), and eventually a federated model, where responsibilities are shared between central and business units.
For data: Assess your organization's readiness across skill sets, use case suitability, tools, and data quality. Make centralized decisions where needed.
For buy-in: Identify relevant stakeholders, preferably senior leaders, and equip them with solid business cases that include direct and indirect benefits (innovation, customer experience, revenue, NPS, cycle time reduction, etc.).
Monitoring: Establish periodic mechanisms to monitor adoption, challenges, and ROI.
Scott Rayburn: Great, Mayank. Please keep going. I just wanted to note that Yuri Gubin is not feeling well, so he had to drop off, but we still have Mayank and me to answer any of your questions, and we'll finish the presentation.
Mayank Maria: Great, Scott. This is really the last view I want to share. We discussed the challenges, and we're seeing that service provider partners are rising to the occasion and helping enterprises figure out many of these challenges. They're building advisory capabilities to facilitate decisions around prioritizing use cases, technology suitability, data readiness, governance, and so on. They provide skilled talent around different technologies and data initiatives. The service provider world is doing a commendable job scaling up talent around new tools and technologies, staying relevant for enterprises, and building their own IP and frameworks that come in handy as enterprises embark on this journey. Service providers are becoming an indispensable part of the puzzle, helping solve challenges and realize benefits from these initiatives.
I'll pause here, Scott, and invite you in.
Scott Rayburn: Thank you. Next slide, please. As they say, the show must go on. That last slide was relevant because it talks about the relevance of solution providers such as DataArt in this ecosystem. If you think of AI as a gold rush, we're like the picks and shovels – the people on the ground who help enterprises digitally transform. We've been doing this since 1997, riding every wave of technology change and helping customers get into the digital future again and again. We're much larger than we were back then – over 5,000 people across 30 locations. We've got the scale to support any transformation project.
Our clients include 400+ leading companies worldwide. Our motto is "partners for progress." We're here for long-term transformation, whether that's RPA, traditional AI, or generative AI. We can walk with you across this whole landscape.
Our AI lab is an internal investment that is available for clients to add value to engagements. We specialize in NLP, gen AI, predictive systems, computer vision, model and algorithm development, and data science. We have proprietary frameworks to get to value faster.
We also offer AI consulting, starting with rapid assessment, moving to strategic development, and ending with implementation. We don't like to get too far ahead with this offering because technology changes every 6 to 9 months. Roadmaps can aim for five years, but things will change in that time.
Last year, we developed over 40 POCs, and that number has certainly grown in the first half of this year. Over 20 of these have been for real clients, and we're on this journey with them. For example, we built a generative AI chatbot for a leading European airport for customer support. It used a RAG solution accelerator, which increased infrastructure development speed by 90%. The customer experience was fantastic compared to what they had before, where duty managers had to manually deal with an overwhelming workload of chats, leading to delayed response times. This is a classic case for a generative AI chatbot. We did this in partnership with AWS, and we can do it with any partner, model, or technology stack.
Now, let's take questions from the audience. Please put your questions in the chat. To start, I have a question of my own. We've focused a lot on where you should deploy AI or process optimization, but which areas of the business are the worst fit for adopting AI? What processes can't be optimized? Where are we having the most trouble? Mayank?
Mayank Maria: A quick answer, Scott: areas that could be addressed with something as simple as RPA shouldn't be complicated by implementing AI. If the process is structured and rule-based, or if a simple Q&A suffices, bringing in a conversational AI agent may do more harm than good to the customer experience. Unnecessary complications with AI should be avoided. I can't generalize by business process, as there are nuances, but broadly, avoid unnecessary complications.
Scott Rayburn: Got it. It's not a magic bullet. Sometimes the best solution is the simplest one.
Any recommended reading on this topic? Mayank, as an analyst, where do you get your information?
Mayank Maria: We follow leading companies at the forefront of generative AI, including toolmakers and hyperscalers. We have many conversations with enterprises to understand their challenges and priorities, and we learn from companies like yours, Scott, who are close to enterprise needs. As analysts, we package all this information to make it consumable. We have a lot of published research on AI, generative AI, and RPA, and we've partnered with companies like DataArt on initiatives and our own publishing.
Scott Rayburn: Thank you. We have a question from the audience: What about investment protection for the enterprise? If a company invests in solutions or integrations with current AI APIs, what is the outlook for this being a good investment in a few years?
Mayank Maria: It depends on scale. If your implementation isn't massive, it makes sense to leverage third-party solutions, especially from hyperscalers. However, as scale increases, the cost of using ready-to-use AI packages may become higher than building your own. You also need to consider the market's conviction in the solution you choose, how current and relevant it will remain, regular updates, and ongoing investments. These factors are critical in deciding whether to buy or build.
Scott Rayburn: It's the timeless question of future-proofing. Is it possible, and at what cost? We can help, but it's always a challenge. It looks like there are no more questions.
I'd like to thank our speakers today, Mayank from Everest Group, Yuri, who had to leave early, and our audience. Thank you all very much. If you have any questions, check out our contact page or email us directly. Thank you for your time, and have a fantastic day.
Mayank Maria: Bye. Thanks, everyone. Take care.
