A practical approach to getting started with generative AI
A practical approach to getting started with generative AI
A practical approach to getting started with generative AI
As generative AI (gen AI) continues to garner massive attention, companies are scrambling to make sense of this powerful next step in AI’s evolution. As they do so, executives need, at a base level, to understand why gen AI is so different, where in the business it can have the most significant impact; and what companies should do to capitalize on its potential fully.
Why is gen AI different?
Plainly put, genAI holds tremendous significance. Why? It’s because the technology is so different from, and so much more powerful than, what we call “traditional” AI.
One of the biggest obstacles companies historically have faced in getting significant benefits from traditional AI has always been figuring out how to tap into the massive caches of unstructured data across the enterprise. GenAI offers an unparalleled ability to harness such data in ways traditional AI has struggled to do. These untapped data sources are now accessible to drive increased efficiency, uncover previously unknown insights and, overall, an improved business outcome.
Although gen AI primarily focuses on predicting the next token, it can also excel in various tasks such as information retrieval, classification, and summarization, ultimately enhancing automation with greater efficiency and effectiveness.
Additionally, unlike traditional AI, gen AI excels in its ability to learn through reinforcement via either unsupervised or semi-supervised human feedback. This approach enables improvements in model performance and higher accuracy overall.
In short, these characteristics alone make gen AI a true force to be reckoned with and a potential source of major increases in a company’s productivity, efficiency, and innovation.
Companies must address many important factors to maximize gen AI’s potential.
Where can the technology have the biggest impact?
In addition to trying to get a handle on why genAI is such a significant step forward, executives are also dealing with a very practical issue: how and where to deploy the technology for the greatest benefit.
Historically, companies found it challenging to scale traditional AI at an enterprise level as the technology was often costly to deploy for a certain set of use cases, and the business case of rolling those out rarely justified the effort. That’s not the case with genAI. Its versatility of applicability across multiple use cases enables companies to solve larger, more expansive problems and, consequently, create value at scale. Organizations will likely start utilizing gen AI in areas with the lowest risk, ensuring we witness value coming from internal applications before customer- or supplier-facing ones.
Consider human resources. In the past, employees who had complex questions about their benefits had to have detailed conversations with a benefits administrator. This took a lot of time and effort, especially on the administrator’s part. With gen AI, a virtual assistant can sift through the complex unstructured knowledge repository of benefits and policy documents and accurately answer employees’ questions within seconds. And, unlike human administrators, that assistant can serve all employees at scale without being overwhelmed or overworked. Gen AI also could dramatically streamline finance and accounting. Today, chief financial officers create a wide array of dashboards they must painfully query to generate desired reports, then read and digest these reports to fully understand the company’s financial state. Gen AI can enable the development of a conversational BI interface through which CFOs can ask myriad questions about the company’s financial performance and have the desired information delivered in any format—a report, presentation, or even a video. The time savings would be immense.
Supply chain and logistics, finance and risk management, data analytics and insights, and contact centers are other internal horizontal functions that could similarly be transformed via gen AI. And then there are processes unique to specific industry verticals—such as clinical trials in pharmaceuticals, claims processing in insurance, and equipment maintenance in manufacturing—where gen AI stands to make a massive impact. The applications are virtually limitless.
An organization has to manage its data as a strategic asset that can provide them with immense value.
Why is data so critical for enabling the gen AI journey?
Companies must address many important factors to maximize genAI’s potential. But it all starts with data. To execute the chosen use cases and produce the desired outcomes, gen AI needs access to the right data—whether it’s public or private, structured or unstructured. And that data must be accurate, clean, complete, integrated, and accessible for gen AI to do its job. As the old saying goes, “Garbage in, garbage out,”; it certainly applies to gen AI. One of the unique aspects here involves unstructured data. Given gen AI’s ability to harness unstructured data, it will become just as important as structured data, which historically has been the focus of most traditional data management and analytics efforts. Ensuring gen AI is using the right data and doing so in the right way requires an intense focus on data management, which comprises four distinct elements:
- Data strategy
- Data governance
- Data engineering
- Data operations
Data strategy is the foundation for a company to fuel its gen AI pathway. An organization has to manage its data as a strategic asset that can provide them with immense value. While most enterprises have some handle on managing structured data, unstructured data tends to be very chaotic with documents and images in multiple silos having limited to no governance. Identifying goals and business objectives for utilizing gen AI and enabling an integrated data management system that helps provide easy and secure access to underlying data for building LLMs, which can then be integrated into enterprise applications, is the key to long-term success.
Data governance becomes more complex, as unstructured data and other repositories that have typically been on the periphery of a company’s analytics efforts are now the core focus. Hence, data governance includes data quality management, lineage, and access to unstructured content and knowledge repositories. Establishing data lineage is vital to building trust that gen AI is using data responsibly and ethically. Robust governance provides the guardrails that ensure the tool doesn’t produce inaccurate information, operate with a bias, violate anyone’s privacy, or go astray of the laws (e.g., copyright infringement or stealing intellectual property).
Data engineering is all about how gen AI models use data. Here, a company must identify and bring together the data to train and run the models in offline and online modes. Additionally, there needs to be checks applied to identify which data is off limits due to security and privacy concerns or relevant regulations. While a large part of a company’s established data engineering capabilities and tools can be leveraged, support for ingesting, transforming, and managing unstructured or text data will be required.
Finally, there’s data operations, which involves automating and monitoring gen AI-specific workflows, putting data security capabilities in place, and gradually evolving to support new use cases to scale gen AI’s deployment. Managing these new technologies so that the user experience does not suffer, and creating confidence in the reliability of the solutions has to be a key focus area.
While thinking about operating the data ecosystem effectively, companies must also consider the compute needs for gen AI. Gen AI requires a massive amount of computing power for fine tuning and inferencing, which could put a strain on a company’s existing technology infrastructure. When gen AI is added, chief information and technology officers need to sort through their overall compute strategy and optimal footprint. Lastly, cost management and optimization techniques must be devised and put in place to keep the operating costs under control.
Certainly, companies must consider many other factors before adopting and deploying gen AI. But data and data management are arguably the most foundational to success.
How is human-machine dynamics being redefined?
Gen AI is a beacon of transformation, reshaping the dynamics of the humanmachine relationship. Gen AI extends beyond mere technological innovation or a transient business trend. It represents a profound shift in how humans and machines collaboratively generate novel outcomes.
How does Gen AI contribute to elevating productivity?
Machines with everyday generative capabilities act as supercharged collaborators, enabling individuals to work with heightened creativity and efficiency. The primary focus of everyday gen AI is to enhance and optimize existing tasks, making individuals more proficient in their current roles. Automating routine processes and augmenting decision-making capabilities allows your team to delve into higher-order generative thinking and strategic endeavors.
EXL is uniquely positioned to help companies in any of these situations. Over the past 15 years, we have helped large enterprises pilot and then scale machine learning and AI solutions. Based on our customers’ needs, we help improve time-to-market for digital interventions by delivering solutions that serve business use cases, and deploying accelerators and tools that enable technology agility. We also specialize in working with companies to modernize their data strategy, governance, and operating models to—securely and responsibly—make sense of the immense amounts of information in their complex systems, with a focus on delivering insights and business outcomes. For companies that want comprehensive help, we have developed a cloud-based platform, or workbench, that brings together all the tools, technology, models, and privacy and security that a company needs to deploy gen AI at scale across its organization. This platform also houses EXL’s plug-and-play gen AI solutions that target key functional and industry-specific use cases.
But, successfully using gen AI requires more than just technical capabilities. EXL brings deep knowledge of business processes to help companies apply gen AI in a way that will deliver the biggest business impact. We work with companies to reimagine their processes for gen AI instead of simply applying the technology to execute inefficient processes in a new way. Complementing our process knowledge is our deep expertise in the industries we work in, including insurance, banking, healthcare, retail, and utilities. This enables us to understand the industry-specific nuances and context to consider when developing and deploying a gen AI solution. We work with a wide range of ecosystem partners, enabling us to bring the capabilities needed to a specific initiative to deliver successful outcomes.
Two companies illustrate how we can marshal our capabilities and resources to help companies adopt and benefit from gen AI. The first is an insurance company looking to use gen AI to enhance its claims processing. The company’s fundamental challenge was that claims handlers spent as much as 65% of their time on tedious, manual activities: navigating disparate systems and knowledge management portals to find all the information they needed to resolve a claim. Besides being inefficient, this approach frustrated customers, who wanted a quicker resolution.
In a proof of value, we deployed EXL’s claims solution to assist the gen AI engine, transforming the process in three ways. First, the engine itself extracts all the relevant information on the claim detail, including the claim number, the claim amount, the stage of the claim, and the reserves. Then, it presents on a single screen all the information on a claim, including its stage of resolution, so claims handlers can see everything they need in one place. The gen AI engine can also query various databases for a claims handler engaged in a conversation with a customer who has questions about the claim, delivering answers in seconds. It can generate emails to customers about the state of their claims and even flag indicators of potential fraud or leakage. Al told, the proof of value effort revealed the claims assist gen AI engine could improve the insurer’s claims-handling process by 25% to 30%.
A second company we worked with, a major utility, is deploying EXL’s customer experience agent assist solution to help improve customer contact agents’ interactions with customers calling with an issue. During the call, the solution uses intent recognition, proactive guidance, and real-time “nudges” to give agents the best information to resolve customers’ problems quickly and effectively. After the call, the engine produces a summary and transcription, giving agents easy access to a customer’s service engagement history. We believe this solution will cut the utility’s call handling time in half while boosting call resolution accuracy by 70% to 80%.
Based on our customers’ needs, we help improve time-to-market for digital interventions by delivering solutions that serve business use cases, and deploying accelerators and tools that enable technology agility.
We believe this solution will cut the utility’s call handling time in half while boosting call resolution accuracy by 70% to 80%.
Gen AI’s potential benefits are alluring, which is why every company should at least explore how to use it and the value it could create.
As gen AI garners significant hype and attention, companies are wary of feeling behind and are under pressure to learn and experiment with the technology. This is easier said than done. As with any new technology, gen AI comes with its unknowns and risks, and companies must be careful to avoid rushing too quickly into using a tool at such a nascent stage of its development.
That said, gen AI’s potential benefits are alluring, which is why every company should at least explore how to use it and the value it could create. As part of that exploration, it can help to have an experienced partner with the knowledge of not just the technology and how it works, but also the business and industry acumen needed to deploy GenAI properly for maximum benefit and with minimal risk.
The fact is companies do not have to go it alone with gen AI. EXL is uniquely positioned to navigate the intricacies of gen AI and help companies use it to drive their business forward. Contact us to learn more.
For more information, contact us at https://www.exlservice.com/about/contact-us
Reimagining gen AI
Reimagining gen AI involves envisioning new possibilities, frameworks, and applications for AI technologies that transcend the conventional boundaries of productivity and creativity.
Reimagining gen AI allows organizations to tap into the full potential of artificial intelligence beyond its current applications. It encourages a shift from merely optimizing existing generative processes to exploring new avenues for growth and generative innovation.
Senior Vice President & Chief Digital Officer