Helm CEO Dawood Patel Highlights Intellectual Property Challenges in AI Data Ownership

Dawood Patel, CEO of Helm, has identified a significant intellectual property challenge emerging from the current surge in artificial intelligence adoption. According to Patel, organisations that depend heavily on customer data and digital interactions must reconsider traditional views of technology ownership.

“Many organisations still think about technology in terms of software ownership,” Patel explains. “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.”

AI systems accumulate extensive interaction data over time, including conversations, behaviours, and preferences. This data can evolve into a powerful proprietary asset for businesses. Patel draws a parallel between interpersonal communication and AI interactions, noting that understanding the specific language customers use can foster more meaningful engagements and stronger relationships.

“In many ways, data is becoming the new form of code,” he states. “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, Patel cautions that challenges arise when organisations build AI capabilities on platforms or models they do not fully control. One critical risk is vendor lock-in, where companies invest heavily in platforms that limit their ability to migrate systems, data, or processes elsewhere.

While vendor lock-in is a longstanding issue in software, Patel warns that AI systems amplify this 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,” Patel says. “What initially looks like a technology decision can later become a financial one.”

Over time, migrating away from such platforms may require abandoning years of investment in integrations, automation flows, and accumulated datasets. At that stage, the decision to move transcends technology considerations and becomes a financial discussion involving the CFO, due to potential write-offs of significant investments.

The rapid adoption of frontier AI models, including large language models and AI assistants, adds further complexity to intellectual property concerns. While these tools offer powerful capabilities and accelerate development, they compel organisations to scrutinise 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,” Patel advises. “Even if vendors say information is ring-fenced, organisations need to understand exactly how their data is handled and what rights they retain over the outputs generated by the model.”

Once AI systems become embedded in operational processes, the intellectual property generated through them can become among the most valuable assets a business owns. Patel suggests that organisations mitigate risks by clearly delineating ownership across different components of an AI solution.

Modern AI architectures often comprise multiple layers, including communication channels, AI engines, integrations, and customer data. While technology providers may own the underlying engine or platform, the custom workflows, interaction data, and integrations should remain the property of the organisation deploying the system.

This separation ensures companies maintain control over the intellectual property that differentiates their business. Without upfront discussions on ownership, organisations risk entrenching strategic capabilities within systems that restrict future flexibility.

As AI investments accelerate, Patel recommends leadership teams ask three critical questions before committing to any technology platform:

1. Who owns the data generated by the system? Customer interactions, behavioural insights, and operational data may become one of the organisation’s most valuable assets over time.

2. Can our intellectual property be moved to another platform if needed? Closed architectures may hinder exporting workflows, models, or datasets in the future.