From electric vehicles to medical devices, TE Connectivity’s $16 billion global enterprise is built on scale and innovation. For Chief Data and AI Officer Elena Alikhachkina, that means treating data and AI not as experiments, but as business-critical foundations.

In this final part of the interview with Clyde Gillard, North American AI GTM Leader at HPE, Alikhachkina explores what it takes to move AI from pilots to production, embed responsible practices, and capture real business value.

The first part of the interview covered the company’s approach to data unification and AI-enabled growth.

Why so many AI pilots fail

Across industries, countless AI pilots generate excitement but never make it to production. Alikhachkina believes the root problem lies in how organizations define “production.”

“Everybody has to define what production means,” she says. “For me, it’s about the full cycle — personalization, escalation, the ability to scale. Without that definition, it’s easy to build something narrow that looks promising but can’t grow with the business.”

She draws a parallel to analytics, where models often fail because they are built in isolation. “People say, ‘Let me build a pricing model.’ They pick the best data, define a single source, and stop there. But pricing is connected to products, customers, and market trends. When you look too narrowly, scaling becomes impossible.”

The key, she argues, is to productionalize data around core business entities such as customers, products, and factories. “If your system is built like blocks, it’s easy to adjust and scale. But if every new dimension requires rebuilding the entire system, you’ll hit a wall. For me, data and AI come together — there is no separation.”

Embedding responsibility through literacy

Beyond scaling, Alikhachkina emphasizes the importance of responsible AI. At TE Connectivity, policies governing AI use are paired with a literacy program that equips employees across the organization to engage with AI responsibly.

“The policy summarizes everything, including the responsible use of AI,” she explains. “But the real power is in the literacy program. It helps people understand how to use AI and why it matters.”

She recalls an example shared at the Boston CDOIQ Symposium. “I heard a terrifying story of board members using tools to record board meetings, which they’re not supposed to do. That shows how the literacy gap goes across all levels of organizations.”

Tools like Microsoft Copilot illustrate both the promise and the pitfalls. “It’s a fantastic tool — it helps in daily life. But when companies opened Copilot to SharePoint, they started finding private documents like employee salaries and IP. The issue wasn’t technology. It was behavior. People don’t organize their phones or files, so why would they organize SharePoint? AI forces us to confront that behavioral gap.”

Small successes that signal big shifts

Despite the challenges, Alikhachkina points to several AI success stories that demonstrate impact and scalability.

In healthcare, personalization and next-best-action models have already shown results. “We’ve practiced this with traditional machine learning, and the models perform quite well,” she says. “Personalization is well developed in financial services, too, and we’re seeing strong results in customer engagement.”

Content creation is another area with promise. “Marketing is asking: how can we create better content, aligned with brand and customer messaging? GenAI has a big role to play here. We’re piloting projects that take categorized market research data, layer an agent on top, and share insights across the organization. These are small successes, but they have long-term potential.”

Rethinking data management with AI

For Alikhachkina, AI’s value extends beyond customer-facing applications into the very foundations of data management. She reflects on her earlier work at Johnson & Johnson, where she championed innovative approaches to master data.

“Imagine messy product and customer data. The traditional way is to buy an expensive master data management system — years of work, millions of dollars, and often never successful,” she says. “I saw so many MDM implementations that never ended.”

Instead, she partnered with a Boston-based company to use AI in improving data quality and management. “It was hard to get acceptance at first. Enterprise architects rejected the idea multiple times. But once they saw results, the technology scaled quickly. Using AI to improve data and enable agents to work faster is an area I’ve invested a lot of time in.”

The road ahead

For TE Connectivity, the path forward is clear: build scalable AI by anchoring it in core business entities, embed responsibility through literacy, and apply AI not just at the edges of the business but at its very foundations.

“Data and AI are not separate,” Alikhachkina concludes. “When you design them together, you create systems that are not only scalable, but sustainable for the long run.”

CDO Magazine appreciates Elena Alikhachkina for sharing her insights with our global community.