Embracing the Art of the Possible: A CIO’s GenAI Adventure

As a CIO who has witnessed the transformative power of generative AI (GenAI) firsthand, I can say with honesty that embracing this technology is not an option but a strategic imperative for any modern enterprise. As IT leaders, we try to optimize across the triad of cost, scale, and experience. The beauty is that GenAI has the potential to not only help but also rewrite the rules of what is possible for each.

As an automation industry leader, our experience with GenAI has been akin to surfing a gigantic wave of innovation. Navigating this flood of change has demanded agility, courage, and a proactive mindset. But more importantly, it has opened up a world of opportunities that seemed beyond our reach only one year ago.

As leaders in IT, we can’t afford to wait for the future to arrive; we must create it. If you’re considering your first step into the world of GenAI or seeking insights from someone who’s been down that path, this blog is for you. In this blog series, I intend to share with you our journey of adopting GenAI, the challenges we encountered, and the wins we celebrated. I sincerely hope it can inspire you to capitalize on the incredible opportunity GenAI presents us with.

Chapter 1: The genesis of our generative AI journey at Automation Anywhere

When we started this journey, we were part of the very first wave of enterprises to adopt GenAI. At the outset, there was widespread confusion and concern. Looking back to the beginning of GenAI adoption—it hardly feels like a mere year ago—organizations were understandably skeptical about what is possible with the technology and associated risks. There were questions and uncertainty about what GenAI could achieve for business. In many ways, it echoed the evolution and adoption of cloud technology.

But taking it slow, watching from the sidelines, was not an option for us.

It wasn’t optional for two reasons: our culture and our industry. Trying new things is an ingrained part of the culture at Automation Anywhere. Our dedication to being on the frontlines is formalized within our operations as “Customer Zero,” our program to experiment on ourselves, sharpen our understanding, and empower the development of tried-and-true solutions to bring to our customers. So naturally, when GenAI arrived on the market, we were eager to get started. But that did not prepare us for the enormity of what was happening.

Because, as leaders in Intelligent Automation, the arrival of GenAI represented a disruption to our industry.

Chapter 2: Facing disruption and embracing risk

Disruption is a big word that gets used a lot. But in this case, I can’t think of a better word to capture the seismic shift that GenAI has brought across all sectors. And for us in the Intelligent Automation industry, it was an immediate disruption right out of the gate.

Undeniably, the initial reaction to such disruption is fear. However, we operated under no illusions of a crystal ball that would predict the future trajectory of this technology or the direction the market was heading. We really did not have another option—we had to embody the principle that true courage is not the absence of fear but the ability to press forward despite it.

I’m asked all the time about my perspective as CIO—I’m responsible for governance and enterprise security, so how am I so comfortable deploying GenAI when we don’t have all the answers? It’s crucial to remember that with every revolution in technology—like the pervasive shift brought on by the iPhone in mobile tech—you never have all the controls figured out. The disruption happens first.

The concerns voiced most often, particularly at the onset, revolved around privacy and security. As a CIO, my foremost concern when introducing new technology like GenAI is undoubtedly the security of our data. The protection of sensitive data, whether PII or proprietary business information, from unauthorized access or misuse is paramount.

With GenAI, we knew there were potential threats to enterprise data security. So how did we respond? In the case of data privacy, we decided to work exclusively with public or non-sensitive data until we better understood the implications and could develop concrete solutions.

Our journey was full of bumping into new questions at every turn. And for most of them, we did not yet have satisfactory answers. We had to decide whether we were going to wait for answers to be available or start finding out the answers on our own. Spoiler alert: Looking back, I can tell you that there was a workaround to every problem and every barrier we encountered.

Chapter 3: Resetting our plans and expectations to match the opportunity

After a few months, once we had done more of our own research, we fully realized the size of the potential, the opportunity made possible by GenAI. And so, at this point in our GenAI journey, we reached a stage of realization and reset. The first guiding principle from Navy SEALs to effectively respond to unexpected situations is recognizing the reality of where you are and the situation you find yourself in right now. We realized we had to put aside all of our plans and reset everything. I can personally attest that it’s simply not an exaggeration to say GenAI truly changes everything.

The gloves came off. We had to dig in—with both hands—and dig deep. We examined the building blocks of automation, the very core of what we do as a company and technology platform, to see where GenAI could have an impact and how it could help. And it's probably no surprise that we quickly realized the answer was everywhere.

For example, inputs to automations used to be limited, relying on pre-structured information, with unstructured data requiring pre-training before it could be used for automating business processes. Now, inputs could be anything, data and information in nearly any format imaginable with no pre-training needed. Outputs, too, thanks to GenAI, could take any form. It could be an image or a slide deck. Even the automation itself, the set of rules for the process, could ultimately not need to be defined and programmed—instead, it could be contextually generated, referencing historical data through natural language conversation with GenAI.

GenAI blasted open the existing boundaries of automating.

We began to reassess every element of our business and product trajectory. Existing plans and investments shifted to capitalize on the massive potential and acceleration. New plans meant new OKRs as the reality of this shift spread to the day-to-day activities of our business and product development.

So, too, our tech stack was disrupted. There were tools and technologies we would need, and there were others we would no longer seek to invest in. Which GenAI tools would we want? And what technologies would support the new way of working with GenAI embedded in everyday tasks?

Chapter 4: Embracing the art of the possible

It was April 2023. We’d learned a lot about GenAI. But there were still very significant gaps in our knowledge. It was at this point that we resisted the urge to smooth things over and make assumptions. We could see the beginning of a foundation starting to form, which made for a constant temptation to start building out this foundation. But that really would have been a false path. We could not build a foundation on thin air.

To fill in more of the picture, we knew it was important to discover more, to find out what we didn’t know. And so we began experimenting in earnest.

In fact, we got so into it that we held a company-wide contest, dubbed “Demo Royale,” to harness the energy with healthy competition and find real-life, right-now applications for GenAI that our own employees could uncover and surface using their subject matter expertise. Close to half of our employee base participated in the event. It was very hard to decide winners from the 200+ detailed use cases submitted.

Ultimately, we had to declare a tie for first place, with the winners awarded $10,000 cash prizes. Everyone experimented and played with this technology, and we came out with so many ideas, so much enthusiasm, and a vision for what we could do. The impact was just the tip of the iceberg.

The pilots created during this competition inspired our customer support and IT teams to automate L1 and L2 support for our customers and internal employees end to end. Our globalization team ended up automating language translation for their video content and our marketing team automated the process of creating customer case studies. Our IT organization launched an internal version of an AI assistant, ‘Jarvis.’

Chapter 5: Building our foundation with the right GenAI model

Okay, so how do we make it work? We’d finally arrived at this far-from-trivial question on our GenAI journey. At this point, we started laying a foundation, discovering the building blocks of operationalizing GenAI.

Turning our ideas and use cases into reality meant choosing a team to become GenAI experts. Where should this skill set be held? Which team or business function would be the hub, the home inside the organization for GenAI? Our own CoE, our automation team, was the simplest place to start. It’s a continuing project for the CoE; they are in the trenches every day, hands-on, learning and enriching their GenAI skill set.

Our automation team ramped up fast—and they are absolutely still learning—and we began to embrace the ideas and opportunities coming in from our business teams. We began to actually build these automations with GenAI. I love the example of our contract review automation with GenAI. It’s an automation that saves 4k hours, which is huge. But what I see that is even more important in many ways is how it eliminates frustration between teams and erases friction from the process.

We had to make choices about policy. How do we guide employees about what is and is not okay? Should we allow them to use freely available third-party AI apps? Can they use any AI model they like for internal use? What kind of data can they share while using AI-enabled apps?

And we had arrived at another of the big questions of this journey: which large language model (LLM) should we invest in? For this, we reached out to one of our longtime partners and actually chose two models to go forward with. We also learned the value of training on the right set of data to make the model an extension of our enterprise knowledge.

It is not an exaggeration to say that training and tuning your chosen LLM is the most complex part of this story.

GenAI comes with accuracy and “hallucination” risks, which are more likely to occur when applying GenAI outside of its trained range of expertise. Our experience has shown us the imperative of training your chosen model(s) with your own enterprise data. And yes—test your model. And test again.

I want to come back to the issue of privacy controls because I keep hearing from IT leaders on this. We’ve learned that the complexity of GenAI’s security and privacy concerns arise from its layered structure, including the foundational model, provider, and any third-party components. This is why we chose early on to work with non-sensitive data that have already publicly shared with our community. This was the plan until we could find a way to mask the sensitive data before sharing it with AI prompts.

There was another critical decision we made during the journey: “Our data will never leave the organization.” Some AI model providers are doing a fantastic job creating these models but are too new to be trusted with our data. That meant we'd host the models on our servers so that we could have full control and visibility into who gets access and why.

What we learned—and have continued to develop for you—is that the most effective way for businesses to achieve security, privacy, and accuracy controls is by harnessing GenAI through a secure Intelligent Automation platform. The platform facilitates end-to-end orchestration across systems and users with built-in governance and guardrails, paving the way for the safe and effective use of GenAI.

Excitingly, we’ve made significant strides in embedding GenAI within the Automation Success Platform, offering built-in enterprise security, governance, audit trails, and compliance tools, allowing you to use GenAI securely and create guardrails around its use.

Chapter 6: In the innovation driver’s seat

What I’m excited about right now is that we are in a wide-open innovation race. GenAI has moved the clock forward on the democratization of AI and, thus, automation. How to get things done becomes an AI-first question. Do you remember the first time you saw an automation come together? Once you’ve seen the power and ability of automation, it’s not possible to look at a business process the same way again.

And now, the same thing has happened with GenAI. It’s an AI-first mindset because GenAI makes this available to everyone. Combining the intelligence of AI and the action power of automation is a match made in heaven. And now, gone are the barriers to AI modeling and expensive, complex, expert-time-consuming development. It’s like an all-you-can-AI buffet.

And I won’t go deep into it here, but this re-opens the debate between buying versus building for enterprise technology needs. Building is back on the table—not (yet) to the extent that it will be the default approach, but it can be a real option that might be more economical and scalable, depending on the problem you want to solve. Especially as the need for UI is changing because we can now combine workflows with a GenAI conversational interface instead.

We’re still uncovering so much value. At this point, we’ve invested in working with business counterparts to think about ideas and build our future roadmap.

Leading the GenAI journey

On this frontier, I want to always keep in mind that GenAI is a technology. That means the power and responsibility to harness AI for the organization lies with the office of the CIO. Forecasts anticipate soaring adoption over the next 4-5 years, with 80% of CIOs poised to harness AI.

For me, as CIO, this journey is the adventure of a lifetime.

If you want to lead the GenAI journey for your enterprise, now is the time to face your concerns so you can move forward and lead. I hope our story can support your journey, and I am always thrilled to hear from CIOs and technology leaders on this path. As we uncover more questions—and answers—we’re sharing them with you. So, what’s holding you back right now?