Multi-agent systems for improved decision-making in industrial value chains

This pre-study project aims to establish the foundation for a Swedish ecosystem around industrial generative multi-agent AI systems, supporting the planning of joint efforts and the creation of a strategic roadmap.

The emergence of generative AI Agents, capable of reasoning, decision-making, and autonomous action, is expected to represent a significant breakthrough in 2025. Agents act independently, but within established frameworks that define what they can and cannot do, unlike previous AI systems that functioned as a support system together with a human.

Challenges

Several unresolved questions remain regarding efficient training of the models, agent orchestration and architecture, data flows during use, management of security risks linked to attacks or data leaks, legislation, potential standardization and more. For an efficient industrial implementation these must be addressed.

Project purpose

We aim to identify how a collaborative project with a wide industrial representation can develop knowledge of applying AI agents to support future Swedish industrial leadership.

Expected outcomes

The pre-study will bring together key stakeholders, describe the state of the art and best-practise for industrial applications using AI agents, map industrial needs and perceived obstacles – which will be used for planning joint efforts and a roadmap.