GEMA – Central process control with Camunda
GEMA orchestrates more than 580 cross-team processes with millions of records centrally using Camunda. cimt developed the process control and supported the migration to Google Cloud.
Solution Play · Camunda
Agentic AI with CamundaAI agents inside the process – with guardrails, approvals and full traceability
With agentic AI with Camunda, AI agents do not work next to your processes but inside them: Camunda sets the goal, tools and rules, the agent decides in context – and every step stays visible and controllable.
Platinum
highest Camunda partner level
30 experts
for Camunda and process orchestration
15 certified
Camunda consultants for Camunda 7 and 8
5,000+ PD
Camunda project experience in the last two years alone
Agentic orchestration
The agent works inside the process – not next to it
Many AI agents today run as isolated chatbots or scripts: quick to build, but hard to control. Camunda 8 takes a different approach. With agentic orchestration, AI agents become part of your BPMN processes – user tasks, fixed rules and AI decisions work together in one end-to-end flow.
The agent knows the goal of the process and chooses the right tools to reach it. Camunda decides which tools are available at all, when a person has to approve and which decisions follow fixed rules. This is how agentic AI with Camunda combines the flexibility of AI with the reliability of an orchestrated process.
Building blocks
How agentic AI with Camunda works
AI Agent connector
Connects a language model to the process: goal and context come from Camunda, the agent decides on the next steps.
A toolbox inside the process
Every activity in the ad-hoc sub-process becomes a tool for the agent – from service calls to connectors. Nothing else is available to it.
Rules and people in the loop
DMN decision tables and user tasks can be tools as well – for fixed rules and human approvals where they matter.
Free choice of model
Language models via Amazon Bedrock, Azure OpenAI or Google Vertex AI, among others – in Camunda 8 SaaS or Self-Managed.
Why orchestration
Why AI agents belong inside the process
Traceability
Every step the agent takes is a process step – visible, logged and auditable.
Clear guardrails
The agent only uses the tools the process gives it – nothing more.
Human-in-the-loop
Critical decisions go to a person as a user task – the process waits for the approval.
Rule-based where needed
Whatever follows fixed rules stays in BPMN and DMN. AI is only used where context and language matter.
Use cases
Where agentic AI with Camunda makes the biggest difference
An example from customer service: an agent detects that an SLA deadline is about to be missed, triggers a task in Camunda and adjusts priorities – before a bottleneck occurs.
Incident management
The agent assesses incidents, proposes measures and escalates automatically – Camunda keeps track of deadlines and responsibilities.
IT service processes
Understand and prioritise requests and allocate resources dynamically – with fixed rules for approvals and costs.
Compliance processes
Detect deviations early, check documents and prepare reports – every decision stays documented in the process.
Getting started
From the first use case to an agent in the process
It starts with a concrete process in which decisions need context – for example assessing requests, documents or incidents. Together we define which steps the agent takes over, which rules stay fixed and where a person approves.
Agentic orchestration is part of Camunda 8. If you still run Camunda 7, we show you the right path to Camunda 8; for a quick proof of concept on your stack there is the Process Automation Starter Kit.
Event
Heimspiel KI on 25 November 2026 in Frankfurt
Experience process orchestration and agentic AI live – together with Camunda. The event is held in German.
References
Process orchestration with Camunda in practice
Panasonic Europe – Scalable approval processes with Camunda and DMN
Panasonic Europe digitalises hundreds of approval processes with one generic BPMN model and DMN decision tables in Camunda; cimt is responsible for the methodological design and supports the implementation.
GVL – Data cleansing with Camunda
For GVL, cimt automated data analysis and cleansing with Camunda as part of a new billing procedure and implemented an extensive rule set with BPMN and DMN.
Frequently asked
Frequently asked questions about agentic AI with Camunda
Agentic orchestration embeds AI agents directly in BPMN processes. User tasks, fixed rules and AI decisions work together in one end-to-end flow – the process sets the frame, the agent decides in context.
Yes. Agentic orchestration is part of Camunda 8 and is available as SaaS and Self-Managed. If you use Camunda 7 today, we support your migration to Camunda 8.
The AI Agent connector supports several providers, including Amazon Bedrock, Azure OpenAI and Google Vertex AI. Which model fits depends on the use case, data protection and operating model – we work that out together.
The agent can only use the tools the process provides in the ad-hoc sub-process. Fixed rules stay in DMN, critical steps go to a person as a user task, and every step is traceable in the process.
Camunda plays to its strengths in business-critical, long-running and cross-system processes. For smaller, more technical integrations and quick automations we often use n8n – and the two can be combined.
With an initial conversation and a concrete process in which decisions need context. From there we work out what the agent takes over and which rules and approvals stay fixed.
Contact
Talk to your Camunda partner
Do you have a process in which an AI agent could help? We will get back to you shortly for a no-obligation initial conversation.
Contact person
Your contact for agentic AI with Camunda
Oliver Fromm
Board Member, cimt Group
Note: This text was created with AI assistance.