At the recent VB Transform 2025 summit, top minds from General Motors, Zoom, and IBM gathered to tackle a question almost every big company is facing: Should you use open-source AI models you can tweak and see inside, stick with locked-down proprietary solutions, or find a mix of both?<\/p>\n
For General Motors, the answer often comes down to safety and following industry rules. They\u2019re building technologies that have to work flawlessly in cars, so controlling how their AI behaves is non-negotiable. GM\u2019s approach is all about staying compliant and keeping risk low\u2014a reflection of the realities in the automotive world.<\/p>\n
Zoom takes a different stance. Their focus is on speed and flexibility, especially since their platform needs to adapt to a fast-changing communications landscape. Open models\u2014those that can be customized or built into Zoom’s own tools\u2014fit well here, offering the innovation boost Zoom needs. It\u2019s not just about picking the newest AI for the sake of bragging rights, but figuring out what truly fits their users.<\/p>\n
IBM, with its roots deep in regulated sectors like finance and healthcare, leans toward governance and trust. Here, closed models have clear advantages: more control, stricter data handling, and often a more predictable set of behaviors. But IBM\u2019s team doesn\u2019t knock open models entirely\u2014they recognize that open systems offer transparency and allow outside developers to push innovation at a pace closed systems struggle to match. That kind of openness matters, especially when building trust or moving quickly is the goal.<\/p>\n
Across the board, all three companies agreed: There\u2019s no single right answer. The choice depends on your customers, your data privacy needs, and the specifics of what you\u2019re building. Sometimes a closed, locked-down approach is safer; other times, the ability to collaborate with a community of developers\u2014or customize the guts of your AI\u2014can give you a sharp competitive edge. In fact, many companies are now creating \u201chybrid\u201d setups, mixing open and closed elements to get the best of both worlds.<\/p>\n
One big takeaway from the panel is that AI in the enterprise has moved well past the stage of hype. Companies are now rolling out real, production-ready systems\u2014everything from highly specialized small models to broad, agentic systems that can make decisions and carry out tasks autonomously. As business needs change, so will the AI strategies companies choose. The guiding principle? Stay agile, keep an eye on real business goals, and match your tools to your needs\u2014because what works today could change tomorrow as tech moves on.<\/p>\n