GenAI suffers from data overload, so companies should focus on smaller, specific goals

As generative AI continues to capture imaginations and headlines, it’s also presenting businesses with a complex dilemma: an overload of data without a clear path to practical application. Companies are collecting massive amounts of information, much of it unstructured and stored in disparate locations. But as Chet Kapoor, chairman and CEO of DataStax, noted in a conversation at TechCrunch Disrupt 2024, the answer isn’t just “more data.” Instead, companies need a targeted approach with manageable, incremental goals that can help them unlock value without drowning in complexity.

“There is no AI without data, there is no AI without unstructured data, and there is no AI without unstructured data at scale,” Kapoor said. Yet, despite data’s essential role in fueling AI, he urged caution, advocating for a deliberate focus on product-market fit rather than scaling too quickly.

Don’t Rush—Build for Product-Market Fit

Joining Kapoor at TechCrunch Disrupt were Vanessa Larco, a partner at NEA, and George Fraser, CEO of data integration platform Fivetran. The trio discussed how businesses can develop AI tools that are not only effective but also sustainable. As Kapoor explained, companies often dive into generative AI expecting a single, data-driven solution that will revolutionize their operations. Instead, they should concentrate on specific, practical use cases that align with their current capabilities and goals.

In Kapoor’s words, the real magic happens when AI meets actual people who are building solutions on the ground—those pioneering teams who aren’t just reading a manual, but writing it as they go. “The most important thing for generative AI is that it all comes down to the people,” he emphasized. “The SWAT teams that actually go off and build the first few projects—they are not reading a manual; they are writing the manual for how to do generative AI apps.”

Avoiding Data Overwhelm: Start Small and Specific

As any company with data overload can attest, the sheer volume can be paralyzing. Whether it’s structured or unstructured, sensitive or scattered across systems, data complexity is one of the greatest obstacles to meaningful AI. Larco, who works with startups spanning both B2B and B2C markets, recommended an approach that may feel counterintuitive to those who equate generative AI with broad applicability and transformative potential. “Work backwards from what you’re trying to accomplish—what are you trying to solve for, and what is the data that you need?” she advised.

Instead of a wide-ranging AI deployment that attempts to capture the entire organization’s data, Larco suggests identifying specific use cases and using only the data required for those goals. This avoids the pitfalls of “feeding” the AI model with irrelevant or redundant information, which can lead to inaccurate outputs, increased costs, and often, disappointment.

“What we’re seeing is companies starting small, with internal applications, with very specific goals, and then finding the data that matches what they’re trying to accomplish,” Larco added. This focus on narrow, internal applications can help businesses learn to manage AI effectively before expanding to more complex projects.

Solving Problems of Today, Not Hypothetical Tomorrow

With over a decade of experience in data integration, George Fraser of Fivetran knows firsthand the pitfalls of overambitious AI projects. As he put it, businesses often make the mistake of planning for future scalability instead of concentrating on the issues they’re facing today. This emphasis on hypothetical needs can lead to wasted resources on projects that don’t ultimately provide value.

“Only solve the problems you have today; that’s the mantra,” Fraser advised. He noted that the majority of innovation costs—around 99%—stem from things that didn’t work out, not from a lack of preparation for scaling up. According to Fraser, focusing on immediate challenges can prevent costly missteps, helping companies keep both their data and their AI ambitions in check.

Generative AI in the “Angry Birds Era”

Kapoor offered a memorable analogy for the current state of generative AI: he called it the “Angry Birds era.” Much like the early days of mobile apps, today’s generative AI solutions are intriguing, perhaps even useful, but not yet transformative. AI hasn’t yet reached a point where it’s seamlessly integrated into daily life, let alone revolutionizing it. “It’s not completely changing my life; no one’s doing my laundry yet,” he quipped.

However, that doesn’t mean there isn’t significant progress on the horizon. According to Kapoor, companies across sectors are starting to implement small-scale AI solutions—focused, internal applications that allow them to work out the kinks. This phase, he suggested, is essential preparation for the transformative applications expected in the near future. “This year, every enterprise that I work with is putting something into production—small, internal, but putting it into production because they’re actually working out the kinks, on how to form the teams to go and make this happen. Next year is what I call the year of transformation, when people will start doing apps that actually start changing the trajectory of the company that they work for,” Kapoor said.

Looking Ahead: 2025 and Beyond

The path forward for generative AI is one of careful expansion. As companies refine their strategies and begin to see the results of these early, focused efforts, they’ll be better positioned to scale AI projects in ways that genuinely support their broader objectives. By working out the practical details—team structure, data handling, and iterative improvements—today’s AI pioneers are setting the stage for a more meaningful, impactful phase of development in the years to come.

For now, the best advice for companies looking to get started with generative AI is to stay grounded. Focus on solving real problems with specific goals, and don’t be swayed by hype or pressured to scale before the foundation is set. In a field where everyone is learning and every project is a new frontier, those who succeed will likely be the ones who take each step with intention, using only the data they need and always keeping sight of the people who will actually be building—and benefiting from—these powerful new tools.

Generative AI may be promising, but it’s still writing its own instruction manual—and smart companies will be those who recognize that in these early days, sometimes less really is more.