Federation for the agentic era

Why this conversation

Large organisations rarely have one API. They have many services owned by many teams, and now agents that want to call them too. GraphQL federation is one way to put a single, governed graph in front of all of them. Whether it works in practice depends as much on people as on software. At the Nordic APIs Platform Summit in Stockholm the host spoke with Viola Marku, Head of Customer Success at WunderGraph, who works with the company’s enterprise customers.

The story of the summit, including a WunderGraph lightning demo on agents exploring a GraphQL schema, is in Nordic APIs 2026: your next customer is an agent.

GraphQL federation, and where agents come in

WunderGraph, she explains, is a software platform for GraphQL and API management built on federation. It offers software both as open source and under different commercial arrangements, so that organisations can use the power of federation.

The use cases vary. Some of the largest retailers in the world, eBay among them, run their traffic through WunderGraph’s open source components. Smaller companies and startups build on federation for agentic workloads. Asked whether this connects to AI, she said yes: the company is exploring more use cases where the governance that federation provides supports a new way of working, with colleagues who are humans and with agentic workflows that are becoming more common every day.

Adoption is a human problem

Her own work is with WunderGraph’s enterprise customers, and the focus is education: how to promote best practice when working with the stack across a whole organisation. Very often, she says, “it’s often a human problem, not a technology problem.” Teams have to work out ways of working, how to promote their choices and drive adoption, and what motivates people to use the best practice available. The technology and the use cases may look different from one customer to the next, but the human problems are the same.

Protecting creative space

Asked about the future, she sees a paradigm shift driven by AI and a stronger push to innovate. The opportunity is to use it for good and be more creative, while keeping our own cognitive abilities. Running seven streams of work in parallel, each with its own context to hold in mind, can be overwhelming. The balance she describes is to take the operational improvements AI brings, and use the time it frees to think, rather than filling it with more repetitive tasks.

Protecting that creative space, she argues, is key to avoiding burnout and avoiding homogenisation across teams, services and companies. Messaging and look and feel already seem very similar from one organisation to the next, which is an opportunity for those that differentiate themselves. Feeling more efficient and productive every day is an attractive line and a dangerous one: without protected time, organisations lose the edge of bringing something new, which sometimes means spending hours on something that is not efficient on paper.

An automotive use case, end to end

Beyond retail, she described an end-to-end deployment for a well-known car manufacturer, from pre-sale to post-sale: marketing, pre-ordering, processing, back-end logistics, delivery and maintenance. Each is a vertical slice of the whole experience. WunderGraph provides the structure to run the services, but customers still decide their own approach; its architects advise on best practice from what they have seen in the industry, with people who have worked in the field for many years. The aim, she says, is to sit down together, understand the pain points and where a customer wants to go, and be partners in that.

To get in touch, she points to wundergraph.com and to LinkedIn.