Chet Kapoor is the chairman and CEO of DataStax, a data management company. He said, “There is no AI, without data. There is no AI, without unstructured information. And there is no AI, without unstructured information at scale.”
Kapoor kicked off a discussion at TechCrunch Disrupt 2024 on “new data pipes” in the context modern AI applications. He was joined by Vanessa Larco – partner at VC firm NEA – and George Fraser – CEO of data integration platform Fivetran. The chat covered a wide range of topics, including the importance and role of real-time information in generative AI. However, one of the most important takeaways from the conversation was that product-market fit should be prioritized over scale, especially in the early days of AI. It’s simple advice for companies who want to dive into the world of generative artificial intelligence: don’t get too ambitious at first and focus on incremental, practical progress. The reason? We’re still trying to figure it all out. Kapoor stated that the most important aspect of generative AI is its people. The SWAT teams who actually build the first few apps — they’re not reading a guide; they’re writing the manual on how to do generative AI applications. Larco, a B2C and B2B startup expert who sits on many boards, suggests a simple but pragmatic approach to unlocking value in the early days.
Larco said, “Work backwards to see what you want to achieve — what problem are you trying solve and what data do you need?” “Find the data, wherever it is, and use it for this.”
This contrasts with trying to implement generative AI throughout the company, throwing all the data at the large-language model (LLM), and hoping it will come up with the right answer at the end. Larco says that this will create an expensive and inaccurate mess. She said, “Start small.” “What we see is that companies start small, with internal apps, with very specific goal, and then find the data that matches their goals.”
Fraser has led the “data movement” platform Fivetran for 12 years, gaining big-name clients such as OpenAI, and Salesforce along the way. He suggested that companies focus on the real issues they are facing right now.
Fraser said, “Only solve problems you face today. That’s the mantra.” “The costs of innovation are always 99% the things you built that did not work, and not in things that you wish you planned for scale in advance.” Although we tend to think of these problems in retrospect, they are not the 99% cost that you bear.
Much like the early days on the web, and more recently the smartphone revolution, the early applications and use-cases for generative AI show glimpses of an AI-enabled powerful future. But they haven’t been game-changing.
Kapoor said, “I call this Angry Birds era generative AI.” “It hasn’t changed my life completely, no one is doing my laundry yet. This year, I’ve seen every enterprise I work with putting something in production. It may be small, internal or even just a prototype, but they are putting it in production to iron out the kinks and to figure out how to create the teams that will make this happen. Next year, I’ll call it the year of transformation. People will start creating apps that change the trajectory of their company. He also writes on other topics that he is passionate about, like the business of open-source software. Paul has been covering consumer and enterprise technology for The Next Web, now owned by The Financial Times, and VentureBeat since June 2022. View Bio
