“The purpose of information is not knowledge. It is being able to take the right action”  —  Peter F. Drucker*

What separates agents from apps
Conventional applications execute predefined logic. Agents go further: they pursue goals, select tools and data, sequence steps and initiate actions. The difference is agency. Merriam-Webster defines agency as “the capacity, condition, or state of acting or of exerting power” (Merriam-Webster 2026). In economic theory, agency arises when an agent acts on behalf of a principal within a delegated area of decision-making (Ross 1973, Jensen & Meckling 1976). Delegation creates a control problem: the agent must pursue the principal’s goals within defined permissions and responsibilities. Multiple agents add a coordination problem: their tasks, decisions and interactions must be orchestrated across systems and organisations (see Agents (are back)).

Agency requires coordination and orchestration
An agent may decide what information it needs, which service to call and which agent to involve next. Chained interactions can therefore cross application, company and jurisdictional boundaries. This raises practical questions:
     • Is each agent and counterparty authentic?
     • Which data, tools and services may they discover and use?
     • Which access and usage policies apply?
     • How are tasks sequenced and conflicts resolved?
     • Can decisions and actions be traced and attributed?
These challenges resemble those already addressed by dataspaces for cross-company data exchange: identity, trust, discovery, access control, governance, provenance, interoperability.

Dataspace superpowers for agents
The IDSA position paper Data Spaces and AI: Trustworthy Agentic Participation in Data Spaces shows how established dataspace mechanisms can support agentic workflows rather than being reinvented in a separate AI stack (Turkmayali, 2026). Dataspaces provide four essential capabilities:
     • Trusted participation: Verify organisations, people, machines and agents
     • Governed access: Control the data and services to be used and under what conditions
     • Discovery, interoperability: Find, combine resources across systems and organisations
     • Provenance and accountability: Trace data sources, interactions, decisions and actions
Agent hubs and dataspaces therefore play complementary roles: Agent hubs coordinate and orchestrate work; dataspaces provide the trusted control plane for participation across organisational boundaries. This combination supports three directions:
     • Dataspaces for AI: Governed access to distributed data and services
     • AI for dataspaces: Improved discovery, automation and data-product creation
     • Agentic participation: Autonomous interaction within established governance boundaries

What this means for business leaders
Agentic AI should not be treated as another isolated application stack. Its value will depend on whether agents can coordinate safely across the data, services and organisations required to complete real work. Three questions should guide development:
     • Where will agents obtain trusted data and services?
     • How will their identities, permissions and actions be governed?
     • Which existing coordination capabilities can be reused rather than rebuilt?
Dataspaces provide a tested starting point. Their superpower is not simply moving data—it is enabling distributed actors to coordinate without surrendering control.

Publications
2-pager by Chris Langdon
Full paper: Turkmayali, A. 2026. Data spaces and AI: Trustworthy agentic participation in data spaces. Position Paper (June), International Data Spaces Association, Dortmund, link

References
Jensen, M. C., and W. H. Meckling. 1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4): 305–360
Ross, S. A. 1973. The economic theory of agency: The principal’s problem. American Economic Review, 63(2): 134–139

* Drucker, P. 1999. Management Challenges for the 21st Century. HarperBusiness: p. 158