Organizational Memory for Agentic Business Process Execution
2026-07-03 • Artificial Intelligence
Artificial IntelligenceMultiagent Systems
AI summaryⓘ
The authors explain that large language model (LLM)-based agents can help automate business tasks but struggle because they don't have access to company-specific knowledge, which is usually scattered across documents like policies and procedures. Instead of embedding this knowledge in individual prompts, which can create isolated and duplicated information, the authors suggest creating a shared organizational memory. This memory would store evolving, company-specific instructions in a way that all agents can use consistently. They also propose a system design for managing and using this memory and test it with a purchasing process example.
LLM-based agentsbusiness process automationorganizational memoryknowledge managementprocess modelsstandard operating proceduresprompt engineeringprocurement processenterprise knowledgeagent architecture
Authors
Lukas Kirchdorfer, Adrian Rebmann, Christian Warmuth, Timotheus Kampik, Theiss Heilker, Gregor Berg
Abstract
LLM-based agents offer new opportunities for automating business process execution beyond the limits of rule-based systems. However, general-purpose LLMs lack the organization-specific knowledge required for reliable execution, which is typically fragmented across human-oriented artifacts such as policies, process models, and standard operating procedures. While such knowledge can technically be encoded in individual prompts or agent-specific retrieval setups, this approach does not scale in enterprises, as it gives rise to knowledge silos and rule duplicates, and makes consistent updates and learning across agents difficult. We argue that this calls for an organizational memory for agentic business process execution: a shared, governed, and agent-consumable reference layer of evolving organization-specific procedural knowledge about how work should be executed. We derive requirements for such a memory, propose an architecture for its curation and consumption, and demonstrate its effectiveness in a proof-of-concept based on a procurement scenario.