BPM Trends

Topic volume over time

Signals per week by canonical topic or cluster.

Cluster radar

• BPM + EA + Process Intelligence Convergence

  ▲Rising

  94 this week · 40 last week · 745 in total

• AI Governance, Regulation & Compliance

  ▲Rising

  55 this week · 28 last week · 738 in total

• Agentic AI & Digital Colleagues

  ▲Rising

  52 this week · 41 last week · 726 in total

• Ecosystem, Industrial Policy & Market Dynamics

  ▲Rising

  3 this week · 1 last week · 69 in total

• Security, Identity & Integration Modernization

  ▼Falling

  3 this week · 6 last week · 121 in total

• Sovereign AI, Cloud & Infrastructure

  ●Stable

  1 this week · 0 last week · 62 in total

Strategic lens · BPM × EA × knowledge graphs

• A Semantic Layers & AI Context▲Rising
• B Process × Knowledge Graphs▲Rising
• C EA × Knowledge Graphs▲Rising
• D Platform & Vendor Moves▲Rising
• E Standards & OSS Ecosystem▲Rising
• F Governed / Regulated Angle▲Rising

Strategic alerts

Moves our agents flag as strategically significant — newest first.

• 10 Oct 2026 · Semantic Layer & AI Context Infrastructure

  SAP HANA Graph vs. Vector Engine Confusion Signals Knowledge Graph Strategy Gap

  SAP's unresolved graph/vector architecture and opaque Knowledge Graph enrichment path create both a risk and opportunity for a portable process/EA semantic core that does not depend on SAP's native KG stabilising in the near term.

• 10 Oct 2026 · Semantic Layer & AI Context Infrastructure

  metaphactory Integrates with SAP HANA Cloud Knowledge Graph Engine on SAP Store

  metaphactory's listing on the SAP Store with HANA Cloud KG Engine integration is a direct federated-access vector for a portable process/EA semantic core — any BPM or EA semantic layer that targets SAP landscapes can now interoperate with this stack without custom ETL.

• 10 Oct 2026 · Process & EA Knowledge Graphs

  Academic: Process Trace Querying via Knowledge Graphs — Bridging Process Mining & Semantic Layer

  Directly validates the portable semantic process core concept: representing event traces as knowledge graphs enables SPARQL/GQL-style querying, pattern detection and conformance checking — foundational primitives for a BT Brain-style process intelligence layer.

• 10 Oct 2026 · Process & EA Knowledge Graphs

  Academic Bridge: Event Knowledge Graphs to OCEL — Comparative Multi-dimensional Process Analysis

  This paper defines the transformation semantics between EKG and OCEL — directly informing the data model and interoperability layer of a portable process semantic core like Business Flows / BT Brain.

• 10 Oct 2026 · Process & EA Knowledge Graphs

  SOUP Tool Simplifies Event Knowledge Graph Construction for Object-Centric Process Mining

  SOUP's no-code EKG construction from event logs directly addresses the usability gap in our portable semantic process core — lowering the engineering cost of populating a process knowledge graph from real event data, which is a prerequisite for any at-scale Business Flows / BT Brain deployment.

• 10 Oct 2026 · Process & EA Knowledge Graphs

  EVErPREP: Event Knowledge Graph Framework for Explainable Process Mining Event Log Preparation

  EVErPREP's EKG-based event log preparation layer is architecturally adjacent to a portable semantic core — it demonstrates how knowledge graphs can serve as the explainability and context infrastructure sitting upstream of process mining algorithms, a pattern directly relevant to Business Flows / BT Brain design.

• 10 Oct 2026 · Process & EA Knowledge Graphs

  Event Knowledge Graphs as Foundational Data Model for Object-Centric Process Mining

  Directly validates the graph-native substrate needed for a portable process semantic core — event knowledge graphs built on LPGs are the structural complement to our Business Flows ontology layer, and Neo4j's accessibility lowers the barrier for OCPM federation scenarios.

• 10 Oct 2026 · Semantic Layer & AI Context Infrastructure

  Lineage-Aware Memory Governance Framework Bridges Semantic Drift and EU AI Act for Enterprise AI Agents

  This paper exposes the critical gap where OSI-style semantic layers lose governance authority the moment an agentic AI generates its own queries — a direct architectural challenge for any portable process/EA semantic core that must remain the authoritative context source for both human BI and autonomous agents.

Topic: Agent & MCP Security Architecture · show all

• 9 Oct 2026 · Agent & MCP Security Architecture · Agentic AI & Digital Colleagues

  Streaming Agents Wrapped in MCP Lose Real-Time Feedback — and How to Fix It (opens in a new tab)

  This technical post identifies a latency and observability gap when wrapping streaming agentic AI tools as MCP endpoints: the parent agent goes silent for 34+ seconds, breaking user trust and orchestration visibility. For BPM/EA architects designing agentic process automation, this highlights a real integration risk when composing multi-agent workflows via MCP — progress signalling must be explicitly engineered.

  Source: Medium — Agentic AI

• 8 Oct 2026 · Agent & MCP Security Architecture · AI Governance, Regulation & Compliance

  Google Research Flags Open Privacy & Security Problems in Agentic AI Architectures (opens in a new tab)

  Google Research scientists identify unresolved privacy and security challenges in agentic AI systems, framed through a 'contextual angle' — likely referencing contextual integrity as a framework for agent behaviour. For enterprise BPM and EA practitioners, this signals that agentic process automation carries inherent governance gaps that lack mature mitigation patterns.

  reported by 3 sources

• 8 Oct 2026 · Agent & MCP Security Architecture · Agentic AI & Digital Colleagues

  Precisely exposes SAP automation and data tools to AI agents via MCP servers (opens in a new tab)

  Precisely has deployed MCP servers across its product portfolio, enabling AI agents to invoke SAP automation workflows, data quality pipelines, and customer communications tools through natural-language interactions without bespoke integrations. For BPM and EA practitioners, this signals a concrete pathway to agentic process execution where AI assistants can trigger and monitor SAP-backed business processes using standardised protocols.

• 8 Oct 2026 · Agent & MCP Security Architecture · Agentic AI & Digital Colleagues

  Explainability Requirements for Multi-Agent Swarms: A Governance Prerequisite (opens in a new tab)

  Multi-agent AI systems introduce structural accountability gaps — decision authority, delegation chains, and action traceability become opaque at scale. For BPM and EA professionals, this signals that agentic process automation requires explainability frameworks before deployment, not after.

  Source: Medium — Agentic AI

• 7 Oct 2026 · Agent & MCP Security Architecture · Security, Identity & Integration Modernization

  Personal Agent Protocol: AI Agents Need Their Own Identity, Not User Impersonation (opens in a new tab)

  This piece argues that AI agents operating on behalf of users should authenticate with their own distinct identity rather than impersonating the user — a key architectural concern for enterprise agentic deployments. For BPM and EA professionals, this highlights a critical governance gap: current enterprise systems assume human actors, forcing agents to masquerade as users, which breaks audit trails, access controls, and accountability models.

  Source: Medium — Agentic AI

• 7 Oct 2026 · Agent & MCP Security Architecture · AI Governance, Regulation & Compliance

  Fail-Closed Control Plane for AI Agents: Policy-as-Code Keeps LLMs from Final Authority (opens in a new tab)

  This article argues that LLMs should only propose actions while a separate policy layer enforces what agents are actually permitted to execute — a fail-closed architecture with human approval gates and policy-as-code. For enterprise BPM and EA teams deploying agentic AI in process automation, this directly addresses the governance gap between autonomous agent capability and operational control.

  Source: Medium — AI Governance

• 7 Oct 2026 · Agent & MCP Security Architecture · AI Governance, Regulation & Compliance

  AI Governance as an Identity Problem: Zero Trust Alone Cannot Govern Agentic AI Behaviour (opens in a new tab)

  The article argues that current AI governance frameworks focus on access control (Zero Trust) but fail to address the deeper problem of AI agents behaving as instructed yet harmfully — a 'doing exactly what it was told' failure mode. For BPM and EA leaders deploying agentic AI in process automation, this highlights a gap between infrastructure-level controls and process-level accountability.

  Source: Medium — AI Governance

• 5 Oct 2026 · Agent & MCP Security Architecture · Security, Identity & Integration Modernization

  API Security Best Practices as AI Agents Become Non-Deterministic System Callers (opens in a new tab)

  As agentic AI systems increasingly act as autonomous API consumers, traditional access control models built around deterministic, human-initiated calls are under stress. Enterprise architects running process transformation programs must rethink identity, authorization, and audit trails for AI agents that invoke internal systems unpredictably.

  Source: Medium — Agentic AI

• 5 Oct 2026 · Agent & MCP Security Architecture · Agentic AI & Digital Colleagues

  MCP and A2A v1.0: Enterprise Agentic AI Architecture for Agent Communication (opens in a new tab)

  0 affect enterprise integration patterns for multi-agent systems. For BPM and EA professionals, the standardization of agent communication protocols is a foundational concern: it determines how agentic AI can be embedded into process orchestration layers and how governance controls can be applied at agent-to-agent handoffs.

  Source: Medium — Agentic AI

• 5 Oct 2026 · Agent & MCP Security Architecture · Agentic AI & Digital Colleagues

  AI Agents as Distributed Systems: Implications for Enterprise Process Control Architecture (opens in a new tab)

  The framing of AI agents as distributed systems rather than monolithic models has significant implications for how enterprises design governance and control frameworks around agentic BPM deployments. Existing distributed systems controls — circuit breakers, idempotency, observability layers — may transfer directly to agentic AI orchestration, reducing the need to build entirely new governance stacks.

  Source: Medium — Agentic AI

• 5 Oct 2026 · Agent & MCP Security Architecture · Agentic AI & Digital Colleagues

  Agentic AI in Finance: Authorization Gaps Risk Systemic Financial Instability (opens in a new tab)

  The convergence of agentic AI with financial services exposes a critical authorization gap: AI agents acting autonomously on financial instructions without robust scope controls could trigger systemic risks akin to bank runs. For BPM and EA professionals, this underscores that agentic process automation in high-stakes domains requires formal authorization frameworks — not just workflow logic — baked into process design.

  Source: Medium — AI Governance

• 3 Oct 2026 · Agent & MCP Security Architecture · Security, Identity & Integration Modernization

  Recorded Future Launches MCP Integration for Agentic Security Intelligence Operations (opens in a new tab)

  Recorded Future's MCP implementation enables agentic AI workflows within security operations, automating IOC enrichment, threat-actor profiling, and write-back to watch lists without human intervention. For BPM and EA professionals, this signals a concrete enterprise pattern where agentic AI closes the loop between analysis and system configuration — a template directly applicable to process intelligence automation.

• 1 Oct 2026 · Agent & MCP Security Architecture · Security, Identity & Integration Modernization

  Salesforce Agentforce Demo Raises Questions Over SAP API Policy Compliance (opens in a new tab)

  A Salesforce keynote demo showing an AI agent executing SAP business processes via Slack has prompted expert scrutiny over whether it breaches SAP's API usage policies, particularly around autonomous LLM-driven access paths. The core tension is whether agent-based access using delegated user authorization constitutes impersonation under SAP's terms — a distinction with major implications for enterprise AI integration governance.

• 30 Sept 2026 · Agent & MCP Security Architecture · AI Governance, Regulation & Compliance

  Okta Blueprint: Govern AI Agent Identities for EU AI Act Compliance (opens in a new tab)

  Okta is positioning identity and access management as a foundational layer for EU AI Act compliance, specifically targeting AI agent governance. For BPM and EA leaders, this signals that agentic process automation requires agent identity lifecycle management — knowing where agents operate, what permissions they hold, and how to audit or revoke access.

  reported by 2 sources

• 30 Sept 2026 · Agent & MCP Security Architecture · Security, Identity & Integration Modernization

  Testing MCP Servers Before Granting Agentic AI Real Enterprise Access (opens in a new tab)

  As enterprises deploy agentic AI systems using the Model Context Protocol (MCP), a structured testing framework for MCP servers becomes a prerequisite before granting agents access to live enterprise systems. For BPM and EA professionals, this signals that agentic process automation rollouts require a new QA discipline — validating tool invocation boundaries, permission scopes, and failure modes before agents interact with production process data.

  Source: Medium — Agentic AI

• 30 Sept 2026 · Agent & MCP Security Architecture · Agentic AI & Digital Colleagues

  Emergent Behaviour in Multi-Agent AI Systems: Hidden Complexity for Enterprise Transformation (opens in a new tab)

  As enterprises scale from single AI agents to networked multi-agent systems, emergent and undesigned behaviours begin to surface — posing governance challenges that no single design decision anticipated. For BPM and EA leaders, this signals that process governance frameworks must extend beyond individual agent controls to encompass inter-agent orchestration and system-level behaviour monitoring.

  Source: Medium — Agentic AI

• 30 Sept 2026 · Agent & MCP Security Architecture · Security, Identity & Integration Modernization

  MCP as a Universal Integration Layer: Implications for AI-Connected Business Systems (opens in a new tab)

  The Model Context Protocol (MCP) is emerging as a potential standard for connecting AI models to enterprise business systems, addressing a critical gap between AI reasoning capability and operational integration. For BPM and EA professionals, this signals a shift toward protocol-driven agentic architectures where AI can traverse process boundaries across ERP, CRM, and workflow platforms.

  Source: Medium — EA

• 29 Sept 2026 · Agent & MCP Security Architecture · Security, Identity & Integration Modernization

  IBM + THG List AI Agent Identity Platform IDTrust on IBM Cloud as 'Know-Your-Agent' Becomes Enterprise Priority (opens in a new tab)

  The Hashgraph Group's IDTrust self-sovereign identity platform is now listed on IBM Cloud Catalog, positioning verifiable AI agent identity ('Know Your Agent' / KYA) as an enterprise-grade capability. For process transformation leaders deploying agentic AI, this signals that agent authentication and authorisation infrastructure is maturing into a procurement-ready category.

• 29 Sept 2026 · Agent & MCP Security Architecture · AI Governance, Regulation & Compliance

  AI Agents on Regulated Financial Data Require Governed Perimeter Data Architecture (opens in a new tab)

  This piece argues that deploying AI agents against regulated financial data demands a governed 'perimeter data' boundary — essentially a data sovereignty and access control layer purpose-built for agentic workloads. For BPM and EA leaders, this signals that agentic process automation in financial services cannot be bolted onto existing data architectures; it requires a deliberate governance perimeter as a prerequisite.

  Source: Medium — EA

• 27 Sept 2026 · Agent & MCP Security Architecture · Agentic AI & Digital Colleagues

  OKF + MCP + RAG: Designing Governed Knowledge Architecture for Autonomous Enterprise Agents (opens in a new tab)

  This article argues that agentic AI failures stem from ungoverned knowledge rather than insufficient model intelligence, positioning Organizational Knowledge Fabric (OKF), Model Context Protocol (MCP), and Retrieval-Augmented Generation (RAG) as complementary pillars of enterprise knowledge architecture. For BPM and EA leaders, this signals that process transformation programs deploying autonomous agents must invest in knowledge governance infrastructure — not just AI capability.

  Source: Medium — EA