The project prepares the rules and technical foundations that public service providers need to interact safely with AI agents acting on behalf of people and businesses. Its focus is on verifying agents’ authority, defining responsibilities, making actions traceable, and enabling cooperation between information systems.
An AI agent is software that can plan steps and use other systems to complete an assigned task. In public services, this could mean finding information, preparing an application or submitting it within the permissions granted by the user.
Without a sovereign trust and governance layer, there is a risk that future public services will become dependent on foreign proprietary solutions or evolve into fragmented, disconnected systems.
For people
In the future, people will be able to authorise their AI agents to interact with public services on their behalf, for example, to apply for permits, submit documents, or communicate with government registers.
For businesses
Business software will be able to interact with government systems automatically and in real time, reducing administrative burden and improve efficiency.
For government
Reason Reserve // Aruait will help the public sector deliver better services more effictively, supporting Estonia's competitiveness at a time when the working-age population is shrinking.
Estonia already has shared infrastructure for accessing public services online:
- Electronic identity (eID) enables people to identify themselves securely online and sign digitally.
- X-tee, Estonia’s data exchange layer based on X-Road technology, enables secure data exchange between organisations’ information systems.
- The state portal eesti.ee provides people and businesses with access to public information and services.
- Pääsuke supports the management and verification of representation and access rights in connected services.
These components provide a foundation for digital interactions. Introducing AI agents requires further agreements about how software may act on someone’s behalf and how each service provider can verify that authority.
Reason Reserve builds on existing capabilities and identifies the additional arrangements needed for agents to work safely across services and organisations.
Today, people often coordinate their own interactions with government: they find the right service, provide information, submit applications and follow up on progress. AI agents could increasingly carry out these steps on their behalf.
For example, an assistant could help prepare a permit application, retrieve information the user has authorised it to access, and submit the application where permitted. This is an illustrative scenario; the project’s pilot will test a selected service process.
Public service providers need to distinguish between three types of activity:
- Accessing information: what information may the agent retrieve, for whom and for what purpose?
- Performing actions: which applications, submissions or changes may the agent make?
- Obtaining approval or a signature: when must a person review, confirm or digitally sign?
Permission to retrieve information should not automatically imply permission to perform an action. Likewise, authorising an agent to carry out a task does not by itself resolve requirements for approval or signing.
The project has four areas of work:
- Agent identification and authority verification. Define how a public service provider can establish whom an agent represents and what it is permitted to access or do.
- Trust arrangements. Identify what information is needed about agents and the organisations or people responsible for them, and how that information can be shared and verified. Assess whether existing solutions can meet these needs and where additions are necessary.
- Interoperability and common requirements. Assess relevant standards and develop guidance and additional requirements for secure cooperation between agents and public services.
- Testing a complete service process. Test agent cooperation, permission checks and evidence of actions within a selected use case. Establish where automation is appropriate and where human intervention, approval or a signature is required.
The detailed architecture and implementation choices will be informed by analysis and testing.
Reason Reserve // Aruait is being developed in collaboration with public sector organisations, research institutions, and technology partners.
Project duration: May 2026 – April 2028
Project budget: 1 million Euros
Funded by: Public Sector Innovation Fund of the Government Office of Estonia
Contact: [email protected]
Project Manager: Anni Lehari, [email protected]
Development Manager: Otto Mättas, [email protected]
- Clear responsibility: actions must be traceable to responsible people and organisations.
- Limited permissions: agents should receive only the rights needed for their assigned tasks.
- Data protection and purpose limitation: access to and use of data must be limited to what is necessary and permitted.
- Revocable authority: permissions must be withdrawable, with further actions under those permissions prevented.
- Traceability: it must be possible to establish what an agent did and under which authority.
- Human control and protection of rights: processes must define when human involvement is required and how to intervene and correct errors.
Related initiatives
Reason Reserve and the eesti.ai AI identity initiative support a shared goal: enabling AI agents to act safely on behalf of people and businesses.
The AI identity initiative focuses on the user’s side of delegation: how a person or business grants an agent authority to act.
Reason Reserve’s main focus is on the public service provider’s side: how an institution verifies that authority, allows appropriate actions, and ensures traceability and clear responsibility.
The initiatives inform each other. Reason Reserve also examines how existing digital identity, data exchange and authorisation services can support these interactions.
Collaboration
Reason Reserve brings together public authorities, researchers and technology partners. All paid work commissioned under the project is procured through transparent procedures in accordance with applicable procurement requirements.
Organisations and experts interested in contributing can register their interest.
Market consultation
The date and participation details for the planned market consultation will be published on this page once confirmed.
Project Launch Event 12th June 2026
During the event, we discussed:
- the activities planned for the project;
- the technological and strategic objectives we aim to achieve; and
- how organisations and experts can contribute as project partners.
Event materials
Explore the presentation, summary, and other materials from the information session below.
The Aruait project is co-funded by the Cohesion Policy Funds Operational Programme for 2021-2027 under the measure 'Enhancing the Innovation Capacity of the Public Sector.'
Related documents
Last updated: 21.10.2026
