Mind over machine

What if Al didn't just follow rules but wrote them, watched them, and called for backup when needed? This isn't automation 2.0, it's a thinking ecosystem where agents act, adapt, and align. Welcome to the future of enterprise intelligence

- Ashok Pandey

In the evolving world of artificial intelligence (AI), buzzwords come and go, but platforms that deliver measurable outcomes stick around. For Jan Wuppermann, SVP & Head of Data & AI, APAC, NTT DATA, the real game-changer isn't just generative Al or large language models, it's how modular Al agents are being orchestrated to solve real-world enterprise problems across industries. In this conversation, he shares the architectural vision behind NTT DATA’s Smart Agent Ecosystem, the role of Al governance, and how agentic intelligence is transforming the relationship between traditional automation and emerging Al layers.

From cloud-native to cloud-agnostic

The platform’s origin lies in a strategic partnership with Microsoft Azure, which provided a foundation for early tooling and co-creation. But as the ecosystem matured, the architecture evolved into a factory-like model, capable of creating, running, and monitoring Al agents across environments, from Azure to GCP, AWS Bedrock, and even private Al instances.

This shift from single-cloud dependency to cloud-agnostic flexibility was critical. Interoperability, once a challenge, is now a baseline expectation. Hyperscalers that once built in isolation are now aligning on industry standards, making agent orchestration not only possible but scalable.

This move beyond cloud loyalty isn’t just technical, it’s strategic. It ensures that the ecosystem is ready to scale globally, adapt locally, and integrate rapidly across any tech stack or compliance regime.

APIs as the new glue

The rise of modular agents also brings into focus a crucial piece of the puzzle: how they talk to each other. While the technical depth of agent-to-agent protocols may vary, what anchors the ecosystem is its reliance on standardized APIs. These act as the connective tissue, previously fragmented but increasingly harmonized across hyperscalers.