As artificial intelligence moves rapidly from experimentation into core business operations, organisations are rushing to deploy AI tools across customer service, operations and internal workflows. But in the race to adopt the technology, many companies are overlooking a critical question:
Who actually owns the intellectual property created by these systems?
Dawood Patel, CEO of Helm, says the current wave of AI adoption is creating a new intellectual property challenge for organisations that rely heavily on customer data and digital interaction.
“Many organisations still think about technology in terms of software ownership,” says Patel. “But in AI systems, the real value is often not the software itself. It’s the data generated through interactions and the insights that emerge from it.”
Over time, AI systems accumulate vast amounts of interaction data such as conversations, behaviours and preferences. This information can become a powerful proprietary asset for businesses. We often say that words matter during interpersonal interactions, and this can be applied to AI as well. As an example, just knowing the words your customers utilise in their engagements with your business is a really powerful way of fostering more meaningful interactions and, ultimately, relationships.
“In many ways, data is becoming the new form of code,” Patel explains. “The more interactions your systems process, the more intelligence your organisation develops. That dataset becomes a strategic asset that can improve forecasting, personalisation and decision-making.”
However, problems arise when organisations build these capabilities on platforms or AI models they do not fully control.
One of the most overlooked risks in enterprise AI adoption is vendor lock-in, where companies invest heavily in platforms that make it difficult to migrate their systems, data or processes elsewhere. While vendor lock-in has long existed in software, Patel warns that AI systems amplify the risk because organisations are not just building workflows, they are generating valuable intellectual property through data.
“When companies build complex systems on top of closed platforms, the intellectual property they create can become trapped inside that ecosystem,” he says. “What initially looks like a technology decision can later become a financial one.”
Over time, organisations may discover that migrating away from a platform means abandoning years of investment in integrations, automation flows and accumulated datasets. At that point, the decision to move is no longer purely a technology one, it becomes a CFO conversation because it may require writing off significant investment.
The rapid adoption of frontier AI models such as large language models and AI assistants is adding another layer of complexity to the intellectual property discussion. While these tools offer powerful capabilities and rapid development speed, they also require organisations to think carefully about what data they share with external systems.
“Frontier models are incredibly powerful, but companies need to ask what happens to their data when they use them,” says Patel.
