Architecture of multi-agent AI systems in enterprise
Integrating autonomous agents into enterprise ERP and ECM environments requires strict architectural discipline, using DDD and RBAC to mitigate security risks.
AUTHOR
Serhiy Boyko
Enterprise Systems Architect
Works on practical business-transformation questions, from defining decisions and roles to validating processes, data and team readiness.
Publications
Automating business processes with AI agents: maintaining control
Integrating generative AI into corporate processes using BPMN 2.0 and DMN for secure AI agent orchestration and maintaining system control.
Secure process automation using AI agents: risk mitigation architecture
How to integrate autonomous AI agents into enterprise systems, avoid the risk of Excessive Agency, and maintain full control using BPMN 2.0 orchestration and DMN rules.
Integrating AI/ML into products: From prototype to production
Transitioning from experimental ML models to robust enterprise systems using AWS Well-Architected Framework, Google SRE, and 12-Factor App principles.
Managing technical debt and code quality in enterprise systems
How SRE practices, DORA metrics, and engineering discipline help prevent technical debt, while low-code platforms standardize core architecture.
Pragmatic architecture for enterprise systems: avoiding distributed monoliths
Transitioning to microservices without considering operational complexity creates technical debt. We examine approaches for building stable enterprise systems.
Hybrid development in enterprise: balancing custom code and low-code
How to combine the flexibility of custom development with the speed of low-code while maintaining engineering discipline, version control, and robust system architecture.
AI agents in workflow: first-line processing and BPM future
How to combine the flexibility of AI with the rigid logic of BPMN 2.0 and DMN to automate first-line request processing without security risks or operational chaos.
Process mining: visualizing real flows and shadow processes
Process Mining methodology reconstructs real task flows using digital footprints, revealing hidden bottlenecks and unauthorized workarounds invisible to management.
Automation antipatterns: chaos, RPA, and AI risks
How to avoid cementing inefficiencies, replace fragile RPA scenarios with reliable API integrations, and safely implement AI into operational workflows using BPMN and DMN standards.
Managing automation projects: Process discovery, pilots, and continuous improvement
Why traditional process maps doom IT projects to failure and how to build effective automation based on real logs, BPMN 2.0, and DMN decision tables.
Overview of BPM platforms: process orchestration using BPMN 2.0 and DMN standards
An analysis of architectural approaches to business process automation. Comparing Scriptum, Camunda, Bonita, and UiPath to eliminate shadow processes and enable flexible modeling.