Agentic AI is changing how enterprises build software, run operations, and deliver services. Through platforms such as Hexaware Agentverse™ and Zerovity™, we are helping clients move from AI experimentation to production. But as organizations deploy more agents, a harder question emerges: how do agents built by different teams, on different platforms, find each other, trust each other, and work together across organizational boundaries?
From Powerful Agents To An Open Ecosystem
The next phase of enterprise AI needs more than capable agents working inside a single company. It needs an open ecosystem where agents can discover each other’s capabilities, establish trusted connections, and coordinate work across industries.
That is why Hexaware is actively engaged with Project NANDA, an open-source initiative building the infrastructure for an Internet of AI Agents. At the NANDA Summit in April 2026 at MIT Media Lab, we joined industry and academic leaders to work through the hard problems of real-world deployment: interoperability, security, identity, and the infrastructure that ties them together.
From Conversation To Contribution
Our commitment goes beyond discussion. In September 2026, Hexaware announced its support for the Project NANDA Fellowship, a three-month program in which Hexawarians and external developers contribute directly to open-source infrastructure for the Internet of AI Agents.
Alongside initiatives such as the Hexaware AI Launchpad, the fellowship reflects a simple belief: the next generation of AI infrastructure will be built by industry, academia, and the developer community working together, not by any one of them alone.
Two Focus Areas For Our Nanda Collaboration
Our work with Project NANDA centers on two tracks. Together they answer the questions every enterprise will ask before letting an agent act on its behalf: is this agent right for the job, and is it really who it says it is?
Track 1: Role-Based Agent Organization (RBAO)
The guiding questions: is this agent fit for its role, who should it work with, does it improve over time, and is it still behaving like itself? Track 1 treats agents the way well-run organizations treat people, with clear roles, structured development, and regular review.
- The RBAO Standard defines agent roles, responsibilities, and collaboration patterns in a common way. A setup engine then uses those definitions to assemble agents into teams suited to the task.
- The Agent Lifecycle (AELC) guides how an agent is built, deployed, and improved, so that progress is deliberate and measurable rather than ad hoc.
- Holistic evaluation checks an agent against its role across capability, safety, and consistency, and keeps checking to catch drift before it becomes a problem.
Track 2: Identity Continuity
The guiding questions: is this agent who it claims to be, and is it legitimately still the same agent? As agents are updated, retrained, and moved between platforms, trust depends on identity that holds up over time.
- Agent discovery. Agents need a reliable way to find each other by capability, across registries and organizations.
- Agent registry. A capability passport gives each agent a verifiable record of what it is, what it can do, and who stands behind it.
- Identity continuity. Agents need a way to prove they remain the same legitimate agent through upgrades, model changes, and handoffs, so trust earned once is not lost or spoofed.
Discovery is where these tracks meet the wider ecosystem, and it is where NandaIndex comes in.
NandaIndex: One Resolution Layer, Many Platforms
NandaIndex is an important outcome of this broader effort. Rather than creating yet another isolated agent registry, it provides a common resolution layer that connects the discovery systems already in use, building on open approaches such as AI Catalog and Agentic Resource Discovery (ARD).
That design matters. Enterprises keep their preferred platforms, gateways, and security controls. Smaller businesses and individual developers can make their agents discoverable without running complex infrastructure. And organizations with mixed technology estates get discovery that works across all of them, not just one.
Combined with identity verification, authorization, and governance, this separation of discovery from the underlying platform lays the foundation for secure collaboration between agents across organizations.
“For the agentic economy to scale, AI agents need cross-platform interoperability. At Hexaware, our collaboration with the open-source Project NANDA reflects that commitment, and NandaIndex is a practical step toward making AI agents accessible, safe, and useful across industries, organizations, and communities.” Erik Haug, Global Head of AI Partnerships, Hexaware
Building The Shared Foundation
We see the open agentic web as a natural extension of the connected digital economy. By supporting Project NANDA and NandaIndex, Hexaware is investing not only in enterprise AI capabilities, but in the shared infrastructure and developer community that will let AI agents work together at global scale.
“Enterprises won’t hand real work to agents they can’t place, can’t verify, and can’t hold to a standard. That’s why our NANDA work focuses on two things: making sure every agent is fit for its role, and making sure it is provably who it claims to be, today and after every update. Get those right in the open, and agents from any platform can earn trust the same way. That is the foundation we want to help build.” Satyajith Mundukkal, Global CTO, Hexaware
