Agentic Process Orchestration: Where AI Agents Sit in Enterprise BPM in 2026
Agentic Process Orchestration: Where AI Agents Sit in Enterprise BPM in 2026
For CIOs, Heads of Digital Transformation, and Enterprise Architects evaluating where AI agents fit in an existing BPM stack — and why the orchestration question now matters more than the agent question.
In April 2026, Camunda released version 8.9 with first-class agentic orchestration support: AI agent tasks embedded directly inside BPMN process models, support for the Agent2Agent (A2A) protocol, a built-in Model Context Protocol (MCP) server exposing the orchestration cluster to AI agents, and a user operations audit log covering AI agent activity.
The release is not a curiosity. It marks the point at which AI agent orchestration moved from research-paper territory into the same enterprise BPM stack European institutions have been running for fifteen years.
For European CIOs and enterprise architects, the practical question is not whether to evaluate AI agents. The agents are already in the building — Copilot extensions, LangChain experiments, CrewAI prototypes, n8n flows, vendor-specific assistants from every SaaS provider on the procurement list. The practical question is where the agents sit, what they coordinate with, and who holds the audit trail when one of them acts on production data.
What agentic orchestration actually is
Three distinct ideas often get conflated.
An AI agent is a system component that receives a task and works toward a specified goal, using the tools available to it. It does not choose the goal. It executes the path.
Agentic AI is the broader category — software that exhibits goal-directed reasoning, plans steps, calls services, and adjusts to outputs. Agentic AI is not self-directing. The goal is supplied. The autonomy lies in the path between input and output.
Agentic orchestration is what coordinates multiple agents — and multiple non-agent systems — across an end-to-end process. The orchestration layer is where the BPM stack does its work: it knows what the process is, where each step lives, which actor performs it (an AI agent, a human, a service, a script), and how the process audit trail is composed across the trail of execution.
Camunda's positioning of 8.9 — agentic orchestration as the unifying frame — is accurate to the category, not just to the product. Multi-agent automation only becomes operational at scale when the agents are coordinated by something that holds process state, manages exceptions, and produces a verifiable record. That something is a process orchestration layer.
The bot-proliferation parallel
Most European enterprises that started with RPA in the late 2010s know what happens when tactical automation outpaces its orchestration layer. Year one, ten bots, manageable. Year three, 100+ bots running independently across business units, no central process model, no single audit trail, no governance over which bot touches which system at which time.
AI agents are about to repeat the pattern at higher cost.
Three things make the AI agent version worse than the RPA version.
First, AI agent outputs are non-deterministic. An RPA bot executes a deterministic script — when it breaks, the failure is visible. An AI agent produces probabilistic outputs against the same input. Failure modes are subtler and harder to detect.
Second, AI agents now reach across organisational boundaries. The Agent2Agent (A2A) protocol — a real and increasingly adopted open standard for inter-agent communication — means agents from different vendors and domains coordinate without explicit human design. The supply chain of AI decisions extends past the enterprise perimeter.
Third, the regulatory horizon has changed. The EU Digital Operational Resilience Act (DORA) requires documented end-to-end resilience for financial entities. Sector-specific compliance frameworks — MaRisk for German banking, DNB guidelines for Dutch banks, GxP for life sciences — increasingly expect a traceable record of automated decisions. Ungoverned AI agents create regulatory exposure that scales non-linearly with the agent count.
The orchestration layer is what prevents the second bot-proliferation cycle. It defines the process the agents are executing inside, the steps that remain deterministic, the steps where AI judgement is allowed, the human-in-the-loop checkpoints, and the audit trail that proves how each instance of the process actually ran.
The orchestration requirements in 2026
A workable agentic orchestration layer has to do four things, all at once.
It has to model the process in a standard notation. BPMN 2.0 is the standard. Process models that already exist in ARIS, Signavio, BIC Platform, or Camunda 7 are the starting point — not an obstacle to be migrated.
It has to execute the model on a maintained engine. The engine has to run deterministic process logic for the predictable steps and delegate to AI agents — or human tasks, or service tasks — where dynamic behaviour is warranted.
It has to maintain a single audit trail. Every step — agent decisions, human approvals, system calls, exceptions — has to be recorded against the process instance. For European banking, financial services, energy, and life sciences, this is now a regulatory requirement, not a feature comparison.
It has to coordinate across the stack the enterprise already has. The orchestration layer is on top of, not in place of, the existing modeling tool (ARIS, Signavio), the existing operational systems (SAP, ServiceNow, Temenos, Salesforce, custom Java), and the existing automation layers (Camunda 7, Camunda 8, RPA, low-code platforms).
Camunda 8.9 meets these requirements for enterprises that have already moved to Camunda 8 or are willing to migrate. The release adds first-class BPMN-embedded AI tasks, A2A multi-agent coordination, the MCP server, and the user operations audit log for AI agent activity. For Camunda 8 customers, this is the v8 answer.
For the rest of the European enterprise market — the majority — the question is different.
The vendor-neutral question
Most European enterprises run a mixed stack. ARIS is the modeling system of record. Signavio sits next to it. A Camunda 7 footprint runs in production. SAP Signavio Process Intelligence powers the mining layer. ServiceNow handles IT workflow. The legacy mainframe is held together by 60 RPA bots from UiPath or Automation Anywhere.
A Camunda 8 migration as the answer to the agentic question is, in this context, a multi-year program. For enterprises with significant Camunda 7 production estates and ARIS-modeled process catalogues, the v8 path is not the only option.
DNA's recommendation, across enterprise engagements: keep the modeling tools where they earn their keep, keep the v7 engine where it earns its keep, and add the orchestration layer that connects the AI agents to the existing systems and produces the audit trail. The orchestration layer is the addition. The existing stack is the foundation.
Concordia BPM is DNA's product answer for enterprises on this path. An open AI-powered BPM platform built on a maintained Camunda v7-compatible engine, on a Java/Angular/PostgreSQL stack. The same BPMN 2.0 process models that run on Camunda 7 run on Concordia BPM. The AI capability layered on top — process design assistance, workflow orchestration, monitoring, AI assistant for the process owner — consumes the modeling work that already exists in ARIS or Signavio. The engine is the motor. The orchestration is the car.
For enterprises that prefer to stay on Camunda v7 directly with DNA support, the consultancy also offers a Custom BPA Automation Taskboard and Cockpit built on the Camunda v7 orchestration engine as a proprietary extension. The v7-direct path remains open.
The discovery question for European enterprises in 2026
Where to place AI agents in an existing BPM landscape is a discovery question, not a vendor question.
Three sub-questions matter more than the platform choice.
Where is the process model? If the process catalogue lives in ARIS or Signavio, the AI agents must consume those BPMN models, not require re-modeling. If the catalogue is fragmented across teams, the discovery work comes before the agent work.
Where is the audit trail produced? Regulators do not care which vendor's engine produced the audit log. They care that the audit trail is end-to-end, immutable, and explains how each process instance executed. Orchestration layer choice follows the audit trail requirement.
Where do AI agents add judgement, and where would they create risk? Deterministic process steps — claims routing under fixed rules, vendor onboarding against a checklist, approval workflows with documented thresholds — usually do not need AI agents. AI judgement is most valuable at steps with genuine variance: complex case triage, exception handling, document classification under ambiguity, multi-source decision synthesis.
DNA's discovery-first methodology is built around this sequence. Discovery defines the process catalogue, the audit trail requirement, and the agent placement before the platform conversation begins.
The architecture question, not the agent question
AI agents are not the strategic decision. The orchestration layer they sit inside is.
For European enterprises planning the next phase of automation in 2026, the practical question is which layer of the architecture is missing — and which AI agent investment compounds versus which one creates more governance debt. The agents are coming either way. The choice is whether they are orchestrated.
DNA delivers process orchestration across Camunda 7, Camunda 8, ARIS, Signavio, ServiceNOW, SAP Signavio, and Microsoft Power Platform — and through Concordia BPM where a maintained Camunda v7-compatible engine with an open AI-powered platform layer is the right fit. 75% of DNA's enterprise engagements last 5 or more years, reflecting the multi-year nature of the architectural work the agentic shift now requires.
Book a meeting → https://www.dnaconsulting.io/contact
About DNA Automation Consulting
DNA Automation Consulting is an EU-based BPM consultancy specialising in process discovery, BPMN design, and automation delivery for enterprise clients in banking, insurance, energy, telecom, and manufacturing. 75% of client engagements last five years or more.
Founded by Ante Gudelj and Nikola Dlaka. Clients include Proximus, Equinor, Merck, and NEXI.
Frequently asked questions
For CIOs, Heads of Digital Transformation, and Enterprise Architects evaluating where AI agents fit in an existing BPM stack — and why the orchestration question now matters more than the agent question.
In April 2026, Camunda released version 8.9 with first-class agentic orchestration support: AI agent tasks embedded directly inside BPMN process models, support for the Agent2Agent (A2A) protocol, a built-in Model Context Protocol (MCP) server exposing the orchestration cluster to AI agents, and a user operations audit log covering AI agent activity.
The release is not a curiosity. It marks the point at which AI agent orchestration moved from research-paper territory into the same enterprise BPM stack European institutions have been running for fifteen years.
For European CIOs and enterprise architects, the practical question is not whether to evaluate AI agents. The agents are already in the building — Copilot extensions, LangChain experiments, CrewAI prototypes, n8n flows, vendor-specific assistants from every SaaS provider on the procurement list. The practical question is where the agents sit, what they coordinate with, and who holds the audit trail when one of them acts on production data.
What agentic orchestration actually is
Three distinct ideas often get conflated.
An AI agent is a system component that receives a task and works toward a specified goal, using the tools available to it. It does not choose the goal. It executes the path.
Agentic AI is the broader category — software that exhibits goal-directed reasoning, plans steps, calls services, and adjusts to outputs. Agentic AI is not self-directing. The goal is supplied. The autonomy lies in the path between input and output.
Agentic orchestration is what coordinates multiple agents — and multiple non-agent systems — across an end-to-end process. The orchestration layer is where the BPM stack does its work: it knows what the process is, where each step lives, which actor performs it (an AI agent, a human, a service, a script), and how the process audit trail is composed across the trail of execution.
Camunda's positioning of 8.9 — agentic orchestration as the unifying frame — is accurate to the category, not just to the product. Multi-agent automation only becomes operational at scale when the agents are coordinated by something that holds process state, manages exceptions, and produces a verifiable record. That something is a process orchestration layer.
The bot-proliferation parallel
Most European enterprises that started with RPA in the late 2010s know what happens when tactical automation outpaces its orchestration layer. Year one, ten bots, manageable. Year three, 100+ bots running independently across business units, no central process model, no single audit trail, no governance over which bot touches which system at which time.
AI agents are about to repeat the pattern at higher cost.
Three things make the AI agent version worse than the RPA version.
First, AI agent outputs are non-deterministic. An RPA bot executes a deterministic script — when it breaks, the failure is visible. An AI agent produces probabilistic outputs against the same input. Failure modes are subtler and harder to detect.
Second, AI agents now reach across organisational boundaries. The Agent2Agent (A2A) protocol — a real and increasingly adopted open standard for inter-agent communication — means agents from different vendors and domains coordinate without explicit human design. The supply chain of AI decisions extends past the enterprise perimeter.
Third, the regulatory horizon has changed. The EU Digital Operational Resilience Act (DORA) requires documented end-to-end resilience for financial entities. Sector-specific compliance frameworks — MaRisk for German banking, DNB guidelines for Dutch banks, GxP for life sciences — increasingly expect a traceable record of automated decisions. Ungoverned AI agents create regulatory exposure that scales non-linearly with the agent count.
The orchestration layer is what prevents the second bot-proliferation cycle. It defines the process the agents are executing inside, the steps that remain deterministic, the steps where AI judgement is allowed, the human-in-the-loop checkpoints, and the audit trail that proves how each instance of the process actually ran.
The orchestration requirements in 2026
A workable agentic orchestration layer has to do four things, all at once.
It has to model the process in a standard notation. BPMN 2.0 is the standard. Process models that already exist in ARIS, Signavio, BIC Platform, or Camunda 7 are the starting point — not an obstacle to be migrated.
It has to execute the model on a maintained engine. The engine has to run deterministic process logic for the predictable steps and delegate to AI agents — or human tasks, or service tasks — where dynamic behaviour is warranted.
It has to maintain a single audit trail. Every step — agent decisions, human approvals, system calls, exceptions — has to be recorded against the process instance. For European banking, financial services, energy, and life sciences, this is now a regulatory requirement, not a feature comparison.
It has to coordinate across the stack the enterprise already has. The orchestration layer is on top of, not in place of, the existing modeling tool (ARIS, Signavio), the existing operational systems (SAP, ServiceNow, Temenos, Salesforce, custom Java), and the existing automation layers (Camunda 7, Camunda 8, RPA, low-code platforms).
Camunda 8.9 meets these requirements for enterprises that have already moved to Camunda 8 or are willing to migrate. The release adds first-class BPMN-embedded AI tasks, A2A multi-agent coordination, the MCP server, and the user operations audit log for AI agent activity. For Camunda 8 customers, this is the v8 answer.
For the rest of the European enterprise market — the majority — the question is different.
The vendor-neutral question
Most European enterprises run a mixed stack. ARIS is the modeling system of record. Signavio sits next to it. A Camunda 7 footprint runs in production. SAP Signavio Process Intelligence powers the mining layer. ServiceNow handles IT workflow. The legacy mainframe is held together by 60 RPA bots from UiPath or Automation Anywhere.
A Camunda 8 migration as the answer to the agentic question is, in this context, a multi-year program. For enterprises with significant Camunda 7 production estates and ARIS-modeled process catalogues, the v8 path is not the only option.
DNA's recommendation, across enterprise engagements: keep the modeling tools where they earn their keep, keep the v7 engine where it earns its keep, and add the orchestration layer that connects the AI agents to the existing systems and produces the audit trail. The orchestration layer is the addition. The existing stack is the foundation.
Concordia BPM is DNA's product answer for enterprises on this path. An open AI-powered BPM platform built on a maintained Camunda v7-compatible engine, on a Java/Angular/PostgreSQL stack. The same BPMN 2.0 process models that run on Camunda 7 run on Concordia BPM. The AI capability layered on top — process design assistance, workflow orchestration, monitoring, AI assistant for the process owner — consumes the modeling work that already exists in ARIS or Signavio. The engine is the motor. The orchestration is the car.
For enterprises that prefer to stay on Camunda v7 directly with DNA support, the consultancy also offers a Custom BPA Automation Taskboard and Cockpit built on the Camunda v7 orchestration engine as a proprietary extension. The v7-direct path remains open.
The discovery question for European enterprises in 2026
Where to place AI agents in an existing BPM landscape is a discovery question, not a vendor question.
Three sub-questions matter more than the platform choice.
Where is the process model? If the process catalogue lives in ARIS or Signavio, the AI agents must consume those BPMN models, not require re-modeling. If the catalogue is fragmented across teams, the discovery work comes before the agent work.
Where is the audit trail produced? Regulators do not care which vendor's engine produced the audit log. They care that the audit trail is end-to-end, immutable, and explains how each process instance executed. Orchestration layer choice follows the audit trail requirement.
Where do AI agents add judgement, and where would they create risk? Deterministic process steps — claims routing under fixed rules, vendor onboarding against a checklist, approval workflows with documented thresholds — usually do not need AI agents. AI judgement is most valuable at steps with genuine variance: complex case triage, exception handling, document classification under ambiguity, multi-source decision synthesis.
DNA's discovery-first methodology is built around this sequence. Discovery defines the process catalogue, the audit trail requirement, and the agent placement before the platform conversation begins.
The architecture question, not the agent question
AI agents are not the strategic decision. The orchestration layer they sit inside is.
For European enterprises planning the next phase of automation in 2026, the practical question is which layer of the architecture is missing — and which AI agent investment compounds versus which one creates more governance debt. The agents are coming either way. The choice is whether they are orchestrated.
DNA delivers process orchestration across Camunda 7, Camunda 8, ARIS, Signavio, ServiceNOW, SAP Signavio, and Microsoft Power Platform — and through Concordia BPM where a maintained Camunda v7-compatible engine with an open AI-powered platform layer is the right fit. 75% of DNA's enterprise engagements last 5 or more years, reflecting the multi-year nature of the architectural work the agentic shift now requires.
Book a meeting → https://www.dnaconsulting.io/contact
About DNA Automation Consulting
DNA Automation Consulting is an EU-based BPM consultancy specialising in process discovery, BPMN design, and automation delivery for enterprise clients in banking, insurance, energy, telecom, and manufacturing. 75% of client engagements last five years or more.
Founded by Ante Gudelj and Nikola Dlaka. Clients include Proximus, Equinor, Merck, and NEXI.
Frequently asked questions
What is agentic orchestration?
Agentic orchestration is the coordination of multiple AI agents and non-agent systems (services, human tasks, scripts) across an end-to-end business process. The orchestration layer holds the process model (typically BPMN 2.0), executes deterministic logic for predictable steps, delegates to AI agents for steps requiring dynamic reasoning, manages exceptions, and produces an end-to-end audit trail. It is the layer that turns isolated AI agents into a governed enterprise capability.
How is Camunda 8.9 different from earlier Camunda versions?
Camunda 8.9, released in April 2026, adds first-class agentic orchestration features: BPMN-embedded AI agent tasks, support for the Agent2Agent (A2A) open protocol for multi-vendor agent coordination, a built-in Model Context Protocol (MCP) server enabling AI agents and LLM applications to access Camunda data via a standardised interface, and a user operations audit log covering AI agent activity. The release also adds RDBMS secondary storage (PostgreSQL, Oracle, MariaDB, MySQL, SQL Server, Amazon Aurora) as an alternative to Elasticsearch.
Should European enterprises migrate to Camunda 8 for AI agent orchestration?
For enterprises already on Camunda 8 or planning a v8 migration for other reasons, Camunda 8.9 is the natural answer. For enterprises with significant Camunda 7 production estates, a v8 migration as the answer to the agentic question is a multi-year program. The alternative is to add an orchestration layer on top of the existing v7 engine and the existing modeling tools (ARIS, Signavio) — keeping the institutional investment intact while adding the AI coordination layer.
How does AI agent orchestration relate to RPA?
AI agents and RPA bots sit at similar layers of the automation stack — both execute tasks the enterprise wants automated. The orchestration layer above them is what coordinates them across a process: an RPA bot handles a UI-driven legacy step, an AI agent handles a step requiring document classification or natural-language reasoning, a deterministic service handles a step with fixed rules. The BPM engine orchestrates the path across all three.
What is the role of BPMN 2.0 in agentic orchestration?
BPMN 2.0 (Business Process Model and Notation, the OMG-standardised graphical notation) is the standard format for modeling the end-to-end process. In an agentic orchestration architecture, the BPMN model defines which steps are deterministic, which steps delegate to AI agents, where human approvals sit, and how exceptions route. The BPM engine executes the BPMN. Camunda 8.9, Concordia BPM, and other maintained BPMN engines all run the same BPMN 2.0 standard.
How does DNA approach AI agent orchestration in an existing BPM stack?
DNA's discovery-first methodology starts with the process catalogue, the audit trail requirement, and the AI agent placement question — before the platform conversation. The recommendation depends on the existing stack: Camunda 8 for v8-committed estates, Camunda 7 with a DNA-built BPA Automation Taskboard and Cockpit for v7-direct enterprises, Concordia BPM for enterprises that want a maintained v7-compatible engine with an open AI-powered platform layer, and ServiceNOW / SAP Signavio / Power Automate where those platforms are the natural fit. The orchestration layer choice follows the architecture, not the other way round.
What is agentic orchestration?
Agentic orchestration is the coordination of multiple AI agents and non-agent systems (services, human tasks, scripts) across an end-to-end business process. The orchestration layer holds the process model (typically BPMN 2.0), executes deterministic logic for predictable steps, delegates to AI agents for steps requiring dynamic reasoning, manages exceptions, and produces an end-to-end audit trail. It is the layer that turns isolated AI agents into a governed enterprise capability.
How is Camunda 8.9 different from earlier Camunda versions?
Camunda 8.9, released in April 2026, adds first-class agentic orchestration features: BPMN-embedded AI agent tasks, support for the Agent2Agent (A2A) open protocol for multi-vendor agent coordination, a built-in Model Context Protocol (MCP) server enabling AI agents and LLM applications to access Camunda data via a standardised interface, and a user operations audit log covering AI agent activity. The release also adds RDBMS secondary storage (PostgreSQL, Oracle, MariaDB, MySQL, SQL Server, Amazon Aurora) as an alternative to Elasticsearch.
Should European enterprises migrate to Camunda 8 for AI agent orchestration?
For enterprises already on Camunda 8 or planning a v8 migration for other reasons, Camunda 8.9 is the natural answer. For enterprises with significant Camunda 7 production estates, a v8 migration as the answer to the agentic question is a multi-year program. The alternative is to add an orchestration layer on top of the existing v7 engine and the existing modeling tools (ARIS, Signavio) — keeping the institutional investment intact while adding the AI coordination layer.
How does AI agent orchestration relate to RPA?
AI agents and RPA bots sit at similar layers of the automation stack — both execute tasks the enterprise wants automated. The orchestration layer above them is what coordinates them across a process: an RPA bot handles a UI-driven legacy step, an AI agent handles a step requiring document classification or natural-language reasoning, a deterministic service handles a step with fixed rules. The BPM engine orchestrates the path across all three.
What is the role of BPMN 2.0 in agentic orchestration?
BPMN 2.0 (Business Process Model and Notation, the OMG-standardised graphical notation) is the standard format for modeling the end-to-end process. In an agentic orchestration architecture, the BPMN model defines which steps are deterministic, which steps delegate to AI agents, where human approvals sit, and how exceptions route. The BPM engine executes the BPMN. Camunda 8.9, Concordia BPM, and other maintained BPMN engines all run the same BPMN 2.0 standard.
How does DNA approach AI agent orchestration in an existing BPM stack?
DNA's discovery-first methodology starts with the process catalogue, the audit trail requirement, and the AI agent placement question — before the platform conversation. The recommendation depends on the existing stack: Camunda 8 for v8-committed estates, Camunda 7 with a DNA-built BPA Automation Taskboard and Cockpit for v7-direct enterprises, Concordia BPM for enterprises that want a maintained v7-compatible engine with an open AI-powered platform layer, and ServiceNOW / SAP Signavio / Power Automate where those platforms are the natural fit. The orchestration layer choice follows the architecture, not the other way round.
What is agentic orchestration?
Agentic orchestration is the coordination of multiple AI agents and non-agent systems (services, human tasks, scripts) across an end-to-end business process. The orchestration layer holds the process model (typically BPMN 2.0), executes deterministic logic for predictable steps, delegates to AI agents for steps requiring dynamic reasoning, manages exceptions, and produces an end-to-end audit trail. It is the layer that turns isolated AI agents into a governed enterprise capability.
How is Camunda 8.9 different from earlier Camunda versions?
Camunda 8.9, released in April 2026, adds first-class agentic orchestration features: BPMN-embedded AI agent tasks, support for the Agent2Agent (A2A) open protocol for multi-vendor agent coordination, a built-in Model Context Protocol (MCP) server enabling AI agents and LLM applications to access Camunda data via a standardised interface, and a user operations audit log covering AI agent activity. The release also adds RDBMS secondary storage (PostgreSQL, Oracle, MariaDB, MySQL, SQL Server, Amazon Aurora) as an alternative to Elasticsearch.
Should European enterprises migrate to Camunda 8 for AI agent orchestration?
For enterprises already on Camunda 8 or planning a v8 migration for other reasons, Camunda 8.9 is the natural answer. For enterprises with significant Camunda 7 production estates, a v8 migration as the answer to the agentic question is a multi-year program. The alternative is to add an orchestration layer on top of the existing v7 engine and the existing modeling tools (ARIS, Signavio) — keeping the institutional investment intact while adding the AI coordination layer.
How does AI agent orchestration relate to RPA?
AI agents and RPA bots sit at similar layers of the automation stack — both execute tasks the enterprise wants automated. The orchestration layer above them is what coordinates them across a process: an RPA bot handles a UI-driven legacy step, an AI agent handles a step requiring document classification or natural-language reasoning, a deterministic service handles a step with fixed rules. The BPM engine orchestrates the path across all three.
What is the role of BPMN 2.0 in agentic orchestration?
BPMN 2.0 (Business Process Model and Notation, the OMG-standardised graphical notation) is the standard format for modeling the end-to-end process. In an agentic orchestration architecture, the BPMN model defines which steps are deterministic, which steps delegate to AI agents, where human approvals sit, and how exceptions route. The BPM engine executes the BPMN. Camunda 8.9, Concordia BPM, and other maintained BPMN engines all run the same BPMN 2.0 standard.
How does DNA approach AI agent orchestration in an existing BPM stack?
DNA's discovery-first methodology starts with the process catalogue, the audit trail requirement, and the AI agent placement question — before the platform conversation. The recommendation depends on the existing stack: Camunda 8 for v8-committed estates, Camunda 7 with a DNA-built BPA Automation Taskboard and Cockpit for v7-direct enterprises, Concordia BPM for enterprises that want a maintained v7-compatible engine with an open AI-powered platform layer, and ServiceNOW / SAP Signavio / Power Automate where those platforms are the natural fit. The orchestration layer choice follows the architecture, not the other way round.
Written by:
Nikola Dlaka
Managing Partner
Share with friends: