The shock-and-awe campaign to build generative AI into corporate workflows has created some big challenges for businesses. As many are currently finding out the hard way, it’s one thing to adopt a point solution or a single AI-powered function. It’s quite another to integrate enterprise generative AI solutions that connect across all aspects of a business’s operations and deliver real, sustainable business results.
Increasingly, the biggest obstacle businesses are facing when it comes to implementing this holistic approach to generative AI is data. An enormous amount of data is required to power generative AI applications and—unlike static algorithmic models and earlier versions of AI—these models require real-time data from numerous business functions to unlock their full value. While many businesses have the data and have already piloted generative AI tools that show enormous promise, very few have begun the hard work on data governance and data migration that will be necessary to achieve their longer-term generative AI goals.
Why data matters now more than ever
In fact, according to a recent study we conducted with Forrester, 90% of corporate data and analytics leaders said their business’ overall success relies on consistent, accurate, and fast access to data and insights, and nearly half (49%) said that their most important goal for this year is improving the accuracy of their decision-making through data. These statistics align with the challenges businesses are presently encountering in the development of comprehensive, enterprise-level generative AI strategies.
In the initial stages of predictive analytics, the primary focus was on creating a powerful model, after which companies would input data, activate the model, and await results. This approach evolved with the emergence of AI and machine learning, requiring constant adjustments as models learned from new data. Today, we’re in an era where generative AI functions as a real-time data consumer, demanding continuous updates, ingestion, analysis, and refinement to maintain optimal performance and improvement. Companies that excel in managing their data in this context are poised to lead the AI revolution.
Smart agent assist blazes the trail for enterprise Generative AI
To put this evolution in context, consider the generative AI use case that is currently gaining the most widespread traction inside large enterprises: smart agent assist tools. These solutions, which use generative AI to help customer-facing support personnel develop the next best response in live customer engagement scenarios, are far more sophisticated than the previous generation of decision tree software. The most advanced versions are tracking conversations in real-time, drawing on detailed data from multiple sources about the customer, cross-referencing with company best practices and protocols, and making on-the-fly suggestions to the agent as the conversation continues.
As an illustration, consider a smart agent assist solution developed by my team, working in collaboration with a major UK-based utility company. This solution is able to identify instances where a customer is facing distress or encountering financial challenges. It promptly accesses the customer’s payment history and account details and provides targeted recommendations for assistance programs and strategies that agents can utilize to provide support. All these prompts are served up on the customer service agent’s screen in real-time as the conversation unfolds, drawing on the utility company’s complete knowledge base of customer data, predefined rules, and procedural codes to come up with the best possible response to each customer question and comment.
The addition of the generative AI-powered smart agent assist delivers obvious gains in efficiency by applying a personal, yet consistent approach to interpersonal communication. But it also improves the end customer experience because it puts the agent in a position to help immediately, without having to put the call on hold to scroll through support materials or ask a supervisor for help. Perhaps even more importantly, it also catalogs every single interaction to be analyzed further for areas of possible fine-tuning and improvement.
The power is in the data
Getting to that level of seamless interplay between human customers, human support agents, and AI-powered agent assist technology requires more than just a great algorithm. The most powerful generative AI models in the world are only as good as the data and experts used to train them. The real key to building a smart agent assist solution that’s capable of drawing on data from disparate parts of the organization is strict data governance. If not integrated well, the consistency and integrity of the data in each system will not match up and the solution will not be able to generate useful assistance for the agent.
Amidst the excitement around generative AI, businesses must refrain from hastily adopting single-feature solutions. There has been a tendency among companies scrambling to show customers, investors, and the competition that they are taking a leadership role in generative AI to hastily implement point solutions and proof-of-concept products that do just one thing well. Real enterprise generative AI solutions that will add significant value to the bottom line will be built on the back of robust data management and data governance initiatives.
About the author
Rohit Kapoor is vice chairman and CEO at EXL, a multinational data analytics and digital operations and solutions company.
To learn more, visit us here.
