Ioannis Antonoglou, co-founder and CTO of Reflection AI, told Semafor’s Future Proof that the AI era will force companies to redefine what they actually do and reconstruct their operations using AI as a structural scaffold. He said this transition will take decades, far longer than many in Silicon Valley expect.
Current data collection misses the work itself
Antonoglou argued that businesses today capture the wrong kinds of data. Companies typically record measurable outputs or finished work products, but they often fail to capture the work itself — the hard-to-define causal chains that connect the actions a company takes. Much of the context, he said, is informal: ‘‘we jump on a call, or someone shared something or said something,’’ and the underlying decision processes are not preserved.
A journalism example
He illustrated the problem from journalism: how can a reporter record gut feelings about what might make a good story? How do you capture who a reporter chooses to call first or which questions tend to elicit the most revealing answers? Those subtle elements of work are typically not stored as data today.
New measurement and execution methods are needed
According to Antonoglou, firms must discard decades-old habits and invent new ways to measure and carry out their work with AI in mind. Reflection AI is attempting to help customers with its open-weight models to support that shift. ‘‘It will take maybe a decade or two decades before people really rethink how things are done and how they can just do them with AI in mind,’’ he said.
Uneven industry adoption and market implications
The transition will not occur simultaneously across all industries, which Antonoglou suggested could be beneficial for economic stability. At the same time, uneven adoption creates opportunities for startups that move faster than incumbents to gain advantage.
Corporate move: Nvidia in talks
The Financial Times reported that Nvidia is in advanced talks to acquire Reflection AI. The report did not provide further transactional details; it states only that negotiations are at an advanced stage.
Why this matters
Antonoglou’s argument that companies should focus on capturing workflows and causal links highlights that AI is likely to be more than a tool for speeding existing processes: it may be the foundation for rethinking how businesses operate. Given the long timeframe and differing industry adoption rates, the coming decades could see substantial restructuring and the emergence of new leaders and laggards in various markets.
