DRUID Talks Season 2 Ep#2 - Beyond the Hype: Conversational AI and ChatGPT for the Enterprise World

Season 2, Episode #2 of our DRUID Talks Podcast features Alessandra Capogrosso, Microsoft Data and AI Lead for Central Europe and a published author with over a decade of experience in the tech industry. In this high-energy conversation, we explore the game-changing potential of ChatGPT and its impact on various industries and businesses. Alessandra takes us on a journey through the transformative capabilities of Generative AI, shedding light on industries such as retail, hi-tech, banking, and life sciences, where it's making waves worth trillions. We also discuss the challenges businesses face when implementing ChatGPT in enterprise environments, how to overcome them and the powerful synergy between Microsoft Azure OpenAI and DRUID's Conversational AI platform. Moreover, Alessandra shares her valuable insights on measuring the impact of ChatGPT integration and building a solid business case on how to set clear objectives, track key performance indicators, and balance quantitative metrics with qualitative feedback to unlock the full potential of AI. Last but not least, Alessandra emphasizes the importance of understanding your business needs, fostering a culture of AI adoption, and focusing on people, not just numbers. Let's dive in!

Kieran Gilmurray:

Alessandra is Microsoft Data and AI Lead for Central Europe. She's also a published author who has been in the tech industry for more than a decade. Alessandra has lived, worked and studied in multiple countries around the world and now focuses on the SMB, small, medium enterprise corporate markets and has a passion for data and AI like no other person I have met. Today, we're going to talk about a whole host of AI-related topics, from ChatGPT in the enterprise to data modernization to disconnected AI services to critical project success factors and a whole lot more. So, sit back and enjoy as this is going to be fast-paced, high energy, and there is lots to listen to. Alessandra, welcome today to DRUID Talks!

Alessandra Capogrosso:
Thank you, Kieran! Thanks for having me. 

Kieran Gilmurray:
Well, let's jump straight in here because we have a lot to get through. The past six months have been different to anything that I have seen in the last number of years. They’re probably on the equivalent to the year 2000 or www.com internet boom. But Generative AI and ChatGPT are making huge waves in the industry.

Which companies and industries do you believe will benefit most from this -what I would describe as game-changing technologies?

Alessandra Capogrosso:

Well, I believe, Kieran, before jumping into the answer, we might mention that, if we may want, Generative AI, in general, holds this tremendous promise for various sectors. There is an interesting McKinsey study analysis on the potential of this Generative AI and its impact on the global economy at scale, which is remarkable, which actually estimates ranging from 2.6 trillion to 4.4 trillion annually across 63 use cases today analyzed. And I would note that this technology has the power to revolutionize industries such as retail, to mention a few hi-tech banking and life sciences. In the retail industry, for example, including auto dealerships, Generative AI could create an additional value of around 310 billion only by optimizing functions like marketing and customer interactions. Meanwhile, hi-tech stands on the other hand and benefits the form of generative AI’s capacity to enhance the speed and efficiency of software development, potentially unlocking substantial value. 

In banking and FSI, it has the potential to augment existing AI efforts by taking on lower-value tasks related to risk management, such as reporting or regulatory monitoring and data collection. And these could lead to greater efficiencies and cost savings. And lastly, since I mentioned life sciences, life sciences sector in it, Generative AI shows immense promise and is booming drug discovery, for example, and development processes making notable contributions to advancing medical research and innovation. 

And we can also say that almost 70% of the projected value of this Generative AI use cases is expected to come from four main areas. So we have customer operations, marketing and sales, software engineering and R&D. So these are the top four where we see much of the growth and to make few actionable on customer operations, for example, you may have a customer support team and an e-commerce company using Generative AI to enhance their knowledge management system and employees now can more to ask natural language questions to the AI system. And it responds with relevant information from past interactions with customers, and these enable the team to quickly address customer queries, resolve issues efficiently and provide this personalized upgrade, ultimately improving customer satisfaction and retention. And overall, it has also the potential to bring value not only in specific use cases but also by transforming the internal management system within the organization.
 
These advanced natural language processing capabilities enable in fact to treat and store knowledge by asking questions in a human-like manner to nurture an ongoing dialogue and this is the thing that empowers to access that information swiftly, leading this better decision-making process and leading the development of the company itself. 
 
Kieran Gilmurray:

It's interesting, we're now talking about billions and trillions. I remember back in the day we were talking about just millions, but it shows you the potential of things as well. But it's not all easy, Alessandra. We can’t pretend you can pay $20 a month and everything will just be glorious. 

What are the challenges businesses are seeing using ChatGPT in enterprise environments and how could they overcome these challenges or risk mitigate them?

Alessandra Capogrosso:

I acknowledge that, Kieran. I think the first and foremost it's data security and privacy. Those take the central stage - to ensure the protection of sensitive information and compliance with privacy regulations, robust security measures and meticulous data governance practices are essential and leveraging the secure cloud platform - one of the secure platforms, like Azure, can play a pivotal role in this aspect. 

For instance, a financial institution adopting ChatGPT for the customer separately, for example, 
would benefit from Azure's top encryption and access control to safeguard clients’ financial data, providing an additional layer of security. And this is the very first thing. But then additionally the size of the business it's going to also significantly impact the rollout of a solution such as ChatGPT. Like, larger enterprises with greater resources and either risk appetite can adopt this technology more swiftly. 

Conversely, smaller businesses may adopt a phased approach. They can start with specific departments and gradually scaling up. To mention an example, a tech start-up can implement ChatGPT for automating customer queries in their core product support initially and then expanding to other areas later. And this is the second thing. And lastly, data modernization in general and business education are critical drivers for unlocking the full potential of this solution. You know that organizations need to update their data strategies. They need to make sure that the data is well-organized, and accessible, and that there’s quality data. At the same time, we also note that it's very important to ask the company about AI in general to empower employees to use the technology effectively.  So, a retail chain leveraging ChatGPT for inventory management would focus probably on centralizing and cleaning their data while educating their inventory management team on how to optimize and stock line and strengthening the supply chain.

So it's important to keep the balance of those two things and embrace this concept of digital transformation enabling businesses to be innovative, driving data driven decisions faster and more accurately, I would say.

Kieran Gilmurray:

Well, I really like that answer because there are a couple of things that don't pick up there, Alessandra. The first one is not all, if we call them ChatGPT instances, are secure and therefore organizations actually need to look at what they purchase. And I like the way Microsoft Azure really stands out in that regard. And the second thing is there are no surprises to a degree. The rules of the game are known and therefore all of the stuff you're describing about data security, data practices, good change management, and communication are all available and all recognized, so the companies can put in this technology in a secure and structured way. And I love that you gave us some examples as well. 
If we continue on the examples. DRUID itself has integrated ChatGPT with Azure OpenAI. 

What are the benefits of Microsoft Azure OpenAI platform working with a Conversational AI platform like DRUID in an enterprise environment? 

Alessandra Capogrosso:

Well, it's a good example. First and foremost, I would say it allows businesses to engage in intelligent conversations at scale, around the clock, so ChatGPT, kept with advanced natural language processing capabilities, can efficiently understand and respond to a wide array of queries, enabling organizations to handle larger volume of customer interactions seamlessly. 

And this powerful collaboration also empowers enterprises to create exceptional customer experiences to faster and better responses. And by harnessing the language understanding progress of ChatGPT combined with DRUID’s conversational design and their industry-specific knowledge of business that I think can deliver prompt and accurate and trusty customer inquiries enhancing overall satisfaction and loyalty. And I would mention one particularly exciting advantage is the system’s ability to remember interactions. So connect them and intelligently continue the conversation. 

So by leveraging Azure Open AI in DRUID's Conversational AI applied for the enterprises, the businesses can store this past interaction with users and use the context to apply these amazing, personalized and context-aware responses. And these continuing conversations not only make interactions feel more human-like but also save customers from the inconvenience of repeating the information. 

Kieran Gilmurray:

And I think we've all been there, particularly when it came to the prior incantation of very early Conversational AI, called chatbots. The computer very often, unfortunately, said, “No, didn't remember this” and then we were in that loop. Yeah, but the first thing I find Conversational AI platforms like that really fascinating. Only on the basis that I sometimes think that folks believe that a large language model comes out, does everything in, is just amazing, but it really needs application of purpose to do things; for folks to get companies to really get it to work. 

So, how should companies get started effectively with ChatGPT? You know, I’m not worried, there are challenges and there are fears, but as you described a moment ago, those things can be overcome. 

Alessandra Capogrosso:

Absolutely! There are actually several things that I would say are very much important to consider starting from the very first one; I would say the most important one, which is assessing the use cases where solutions like ChatGPT can make a meaningful impact, and this is the very first stop. So, identify areas in customer support, content generation or knowledge sharing that you would actually benefit from the AI-powered conversational capabilities, laying the foundation for this effective integration. Then, understanding the data requirements because data is the fuel of this technology. 

So, high quality diverse and relevant data is essential for training these models effectively. So I would say companies should have a clear understanding of their existing data and what additional data may be needed to optimize the models and the performance. And to start, I always suggest that a team consider a pilot project that you can start big with a very structured and big project touching the oval rooms of the organization. Or you may want to start small. So considering that pilot project to test ChatGPT's effectiveness in a controlled environment to gain insight. And I think this approach helps understand the impact and the potential challenges before full-scale deployment. So you have this smoother transition, I would say.

And then, of course, you can choose the right model, and you can finetune the model to suit specific business needs, which is also important, but overall, I would say also promoting user awareness and providing training on ChatGPT’s capabilities and potential use cases is also essential. So, educating employees who will be working with the technology enables them to leverage it effectively, enhancing its overall impact in your organization. And, of course, I would mention also that to ensure this return on investment and in part of the solution on the business side define the key performance indicators that are at the core of the conversation and constantly evaluate its effectiveness in achieving defined objectives while testing the solution.

Kieran Gilmurray:

Oh, I adore that answer for so many reasons. I love the top end in the bottom end, in particular, and love the bits in between. But having an actual business purpose is so, so key. Because, Alessandra,
I see too many people getting excited by the technology and just saying we have to have it and use it worldwide, “Where are you going to put it? Don't know, but I need it.” So, the fact that you actually have a business case there and that you can install it and learn, and, importantly, just as you wrapped up at the end there teach people how to use it is essential as well because there's no use having the sharpest knife in the world if you can't actually use it in anger or in some shape or another to make it valuable.

Question and just going back to Conversational AI: Do you think that Conversational AI is pushing 
the adoption of ChatGPT or vice versa? Because large language models, as I mentioned earlier on, don't actually do anything unless they're 
applied in a particular context, surely.

Alessandra Capogrosso: 

I think it’s a symbolic relationship where both Conversational AI and ChatGPT are propelling each other’s rapid adoption, I would say. Conversational AI demand for enhanced language understanding and response generation has led to the growing interest in powerful language models like ChatGPT and Conversational AI platforms and applications such as virtual assistants, chatbots, underscore and significant advanced language capability.

So, specifically, ChatGPT’s ability to comprehend context and generate this human-like response makes it an appealing choice for developers seeking help in their conversation, Conversational AI solutions and deliver better user experience. And at the same time, ChatGPT’s swift adoption is driving the expansion of Conversational AI in general, and as businesses and developers witness the potential of ChatGPT across various domains, and there are, of course, I would say, increasingly inclined to explore and to integrate Conversational AI technologies that leverage this language model.

So there is this interplay, I would say, between Conversational AI and ChatGPT, which creates a positive feedback loop, accelerating the adoption in tandem. 

Kieran Gilmurray:

I like that; one supporting the other. Not the only use case, but I think a very useful one as well. If we go back a little bit as well. Businesses need to get a return from their investments in technology, let's stay on that for a second. But that's hard to identify. In other words, it's difficult to identify an ROI when you maybe don't yet understand the technology when you can't yet see the future, you don't know where it's going to be, and there's a lot of technologies out there that you could potentially invest in.

Alessandra, how would you recommend companies measure the impact of introducing ChatGPT, and how do they go about building that business case right from the very start? 

Alessandra Capogrosso: 

I would say, Kieran, as we mentioned a few minutes ago, they must clearly define their objectives, so they need to set specific goals for the technologies’ integration. Identifying relevant KPIs is essential for tracking and evaluating the technology's effectiveness overall. And to provide even comprehensive understanding of ChatGPT’s impact, collecting baseline data before the implementation is also crucial.

So, this data will serve as a benchmark for comparison and help quantify the improvements achieved post-integration. And in addition to this, quantitative metrics are also vital to tracking user feedback and adoption rates. Positive feedback and high adoption indicate that technology is adding value, and this like being… embraced by the users.

So it's like a balance between what I can measure on a quantitative perspective and what I can track collecting the feedback of the users and my employees, and also measuring time and cost savings achieved, so ChatGPT can demonstrate its practical benefits. So by, for example, automating repetitive tasks, it allows human resources to focus on more strategic objectives, for example. Furthermore, evaluating the impact on customer experience is essential as well, on the other hand.

So if we look on the outside, customer interaction, personalized experiences and reduced resolution times can also indicate improved customer satisfaction, which is also one of the things that we may take into consideration when evaluating actually the overall impact of the solution on our business.

So, I would say, overall, even the alignment of the solution with the organization's strategic initiatives and the contribution to the business growth and the opportunity should be assessed, considering these two things, in my opinion. And by carefully analyzing this insight, these valuable KPIs, the companies can make for sure well-informed decisions about their AI adoption journey that they are embracing. 

Kieran Gilmurray: 

Alessandra, is that true of all sizes of organization, you know, large enterprise, small business, medium-sized?

Alessandra Capogrosso: 

Iwould say absolutely yes. When we mentioned before that you can have different approaches when it comes to how you embrace or adopt a technology according to the size of the company and according to the percentage of risk you want to take, regardless of the fact you are a small business or a big enterprise, as I mentioned these few things are crucial for both in order to embrace this journey. So KPIs, feedback, internal awareness, impact on the customer satisfaction or the customer experience. Those are the things that we need to combine and productivity, the impact on productivity as well. Those are the things that we need to consider when it comes to adopting such kinds of technologies overall.

Kieran Gilmurray: 

It’s funny from your answer a moment ago that we could put a hard financial number on things; so we could put a dollar figure or a euro figure on every aspect of this technology. But that would maybe encourage us to miss some innovation, that would maybe encourage to miss out on some of the solved benefits. In other words, how staff and customers feel or increase productivity.

My question, Alessandra, is: Is there a danger that we try and put numbers on all new innovative technologies and then we miss out on a lot of the benefits that they additionally bring beyond the finance? 
 
Alessandra Capogrosso: 

So I think this is an interesting question, and the reason why it is interesting it's because I agree with you: we can't put numbers on top of everything when we judge the impact of something. That's why I think it's important to consider KPIs and measure KPIs, so we need quantitative measurements to understand what the real impact is on the business. Right? So, this is really important when embracing this kind of strategic decision for sure.

But at the same time, I truly believe, like you said, it's important to evaluate also things that we cannot truly specifically measure, such as the feedback of the employees or the effectiveness of how they're managing tasks, daily tasks right now, thanks to the use of the AI and in general of such smart solutions. And even the feedback and the speed that we are having internally or externally when it comes to productivity is important. So there are things that may be hard to measure, there are things where it's hard to put just a number on them just to quantify them. That's why, overall, when we need to measure the impact or when we need to understand what would be that impact before jumping or embracing such decisions, it's important to try to oversee the full picture—so combining the two aspects. 
  
Kieran Gilmurray: 

I like that answer. I really do.

A final question: If you had one piece of advice to organizations wanting to integrate ChatGPT into their enterprise systems, what would that one piece of advice be? 

Alessandra Capogrosso:

I would say that that would need to be prioritized: an understanding of your specific business needs and use cases before integrating solutions such as ChatGPT. So take the time to identify areas where AI-powered conversational capabilities can make a significant impact, such as customer support, content generation or knowledge sharing. And by defining clear objectives and understanding how these solutions align with your overall business strategy, you can really ensure this seamless and successful integration. And as I mentioned before, even investigating and investing in data is crucial. So, ensure you have relevant and diverse data to train the models effectively and to optimize the performance. So maybe that could be to start, as I mentioned before, choose a pilot project to test that technology first, the effectiveness of the technology in a controlled environment because I think this approach will allow you to assess its impact, gather feedback and address any challenges before a full-scale deployment. I would say remember that successful integration of ChatGPT is not just about technology; it’s also about fostering a culture of AI adoption within your organization. So, educate and train your team to leverage technology effectively and encourage open communication to identify new opportunities for growth and innovation. That's why it's super important to not focus only on numbers, as we said, but focus on people.

Kieran Gilmurray:

Oh, what a wonderful way to end an interview by that last sentence alone. It's focusing on people as well. I think Generative AI and ChatGPT is a really secure technology when you implement the right thing. I think if you do build a business case around it, everything should be done with a business case, but not a sole focus. Alessandra, as we said in the interview on pure metrics, because then you miss out on the people, and you miss out on innovation, and you miss out on a whole host of benefits that make this technology truly wonderful.

Someone said to me recently, “If you can think it and you can describe it or type it, then Generative AI and Open AI can suddenly create it for you.” It's an exciting time in technology once more, and I cannot wait to see what happens in the future. Technologies like Microsoft Azure Open AI and Conversational AI combined together suddenly create a whole host of opportunities for businesses that never existed before. Now, I can't wait to see what we do with these fantastic technologies. Thank you so much for today! That was an absolute pleasure!

Alessandra Capogrosso:

Thank you, Kieran. And if I conclude as well, it just came to my mind, I would add to yours, “It almost seems impossible until it's done”, which is very true. So don't be afraid. Just embrace it. 

 Kieran Gilmurray: 

Fantastic. Thank you so much! 

Alessandra Capogrosso:

Thank you, Kieran!