Can your AI agent prove who it works for? Agion and SIROS set out to find out
Finnish AI company Agion and Sweden’s SIROS Foundation are teaming up to explore how organisations can verify who an AI agent is, who it represents and what it is allowed to do.
Martti Asikainen | 31.10.2026 | Photo by Adobe Stock Photos
Imagine an AI agent placing an order with a supplier on behalf of a company. Before anything ships, the supplier needs to know two things: that the agent really works for that company, and that it is allowed to make the purchase. There is no common, trusted way to check either.
Finnish AI company Agion and the Swedish nonprofit SIROS Foundation want to change that. The two organisations announced a strategic partnership on 28 October to explore how businesses and public bodies can verify an AI agent’s identity, who it represents and what it is permitted to do.
The work is at an early, exploratory stage. The partners have not announced a timetable, products, customers or financial terms.
Why it matters now
As organisations hand AI agents more operational work, questions of authorisation and accountability are getting harder to avoid. When something goes wrong, someone needs to be able to show which agent acted, for whom, and on what authority.
The partnership also lands in the middle of Europe’s digital identity push. EU member states are legally required to offer citizens at least one European Digital Identity Wallet by 24 December 2026, and SIROS’s work grew directly out of European wallet pilots.
“AI should compound your advantage, not your dependency,” says Janne Pulkkinen, Agion’s chief executive and co-founder. In his view, agents add a new layer to familiar questions about who acts on whose behalf. The aim, he says, is to let organisations give agents more autonomy without losing control of their decisions and knowledge.
Leif Johansson, executive director of SIROS, said identity and delegation become more important as agents take on more tasks for organisations.
Agent platform meets identity specialist
Agion was founded in Helsinki in 2025. Its main product, Agion OS, is used to design, deploy and monitor AI agents and AI-driven workflows inside organisations. The platform includes governance controls, access management and audit trails, and can run on infrastructure controlled by the customer.
The company targets public institutions and businesses in sectors including finance, telecommunications and healthcare. According to Agion, Pulkkinen previously led AI projects at Kela, Finland’s Social Insurance Institution.
SIROS, based in Stockholm, develops open-source digital identity, wallet and delegation infrastructure built on international standards. The foundation was set up by Stina Ehrensvärd, co-founder of Yubico, the Swedish company known for its YubiKey security keys.
Its work grew out of cooperation between Yubico and two university networks, Greece’s GUnet and Sweden’s Sunet. That cooperation included wwWallet, an open-source wallet used in European digital identity pilots.
No timetable yet
The partners described the work as exploratory and did not say when results are expected or whether it will lead to new products. It’s worth noting that they are not alone on their quest.
In July, engineers from AWS, OpenAI, Okta and others published an Internet Engineering Task Force draft on how existing standards such as OAuth could be used to authenticate AI agents and authorise their actions.
Agion and SIROS say their work will also build on open standards.
