How to Deploy AI Agents Faster in Business Processes Using Camunda 8.9

Artificial intelligence is the hottest trend today, but for most Ukrainian companies, it still remains at the level of interesting experiments or localized pilots. Everyone understands the potential of AI agents, but when it comes to integrating them into rigorous business processes — with real data, systems, and people — numerous questions arise regarding control and security. This is where the main obstacle to scaling lies.

According to Camunda’s 2026 State of Agentic Orchestration and Automation Report, while 71% of organizations say they are already using AI agents, only 11% have actually put them into production. This means that most companies are still stuck at the proof-of-concept stage. Why is this happening? Because creating AI agents isn’t the hardest part. The real challenge is full-scale orchestration.

Integrity Vision has been working in process automation for over 16 years, and in this article, we’ll explore six key updates in Camunda 8.9. This version provides fast, secure, and reliable orchestration of AI agents, transforming AI experiments into ready-to-use business automation that will help you move from AI pilot projects to the implementation of ready-made solutions in a real production environment.

1. Easy deployment thanks to support for relational databases

Camunda 8.9 now supports relational databases as an additional data store. Now, in addition to Elasticsearch and OpenSearch, Camunda 8 can be deployed using a data storage architecture based on relational DBMSs—a familiar approach for development teams. Support is available for the following databases: PostgreSQL, Oracle, MariaDB, MySQL, Microsoft SQL Server, Amazon Aurora, and H2.

2. Rules and Event Listeners for User Tasks

Camunda 8.9 introduces global event listeners for user tasks. These allow you to register relevant rules at the Camunda 8 cluster level and respond to events within tasks (due dates, SLAs, notifications, validations, and priorities). Analysts and process designers no longer need to configure the same listener for each individual process step.

In any large-scale process, it is critical to monitor tasks performed by people. Deadlines, escalations, and notifications—all of these require reliable oversight. Global event handlers for user tasks ensure that every user interaction with the system adheres to uniform corporate standards. This minimizes the human factor, reduces security risks, and eliminates the need for developers to define the same rules for every step.

3. A complete audit log for every critical action

An enterprise solution can only be trusted when every significant user action is logged and operational transparency is ensured. To this end, Camunda 8.9 introduces a centralized audit log that records who performed an action, what exactly was done, when, and why.

This level of transparency is indispensable for various teams. For example, compliance teams can use the log during audits, operations teams can use it to investigate outages, security teams can use it to detect unauthorized access, and administrators can use it to verify compliance with governance policies.

4. Support for the A2A protocol to build AI agents that can utilize other agents

Today, most companies are moving beyond experiments with individual AI agents toward orchestrating entire ecosystems of agents. According to recent industry reports, over 60% of corporate AI initiatives are shifting toward multi-agent architectures, as real-world business processes are never handled by a single system alone. However, the more agents that are deployed, the more difficult it becomes to manage communication, maintain control, and ensure stability. Without proper orchestration, AI can quickly turn from an advantage into chaos.

Camunda 8.9 introduces dedicated support for the “agent-to-agent” (A2A) protocol. This is an open standard that allows AI agents from different vendors and domains to communicate using structured messages. The A2A protocol includes discovery of the functions available to an agent and the corresponding procedure for exchanging information.

For example, when processing a loan application, an AI agent supporting the decision-making process can independently contact another agent (specializing in fraud detection) and request a risk assessment without creating complex separate integrations. At the same time, thanks to the flexibility of agent configuration in Camunda, it is possible to implement multi-step interactions between agents using information from available knowledge bases, or by involving employees in the process to grant access to sensitive data.

The Camunda platform offers ready-made specialized connectors for this purpose, which significantly accelerate the implementation of multi-agent orchestration while maintaining control, audit trails, and clear process boundaries.

5. Combining Clear Rules with Flexible AI Solutions

Business processes typically operate according to strict, predefined rules, as it is important for companies that all steps follow one another, results are predictable, and every action can be easily tracked. However, in real life, things often don’t go according to plan.

Camunda 8.9 solves this problem using BPMN conditional events (special triggers that respond to changes).

Now, when an AI agent makes a decision, performs an assessment, or makes a recommendation, the system can react instantly—for example, by pausing the current step, redirecting the process, or escalating the matter to higher management if the situation requires it.

As a result, the company gains intelligent automation that handles both routine, standard tasks and any unexpected situations with equal clarity and control.

6. Secure Process Updates Without Disrupting Business

AI-powered workflows are constantly evolving: teams refine prompts for language models (LLMs), add new tools, or modify approval steps. But by the time an update is rolled out, many processes are already running.

Camunda 8.9 allows you to update processes that are already running thanks to support for ad-hoc subprocesses. Now, businesses can migrate active tasks from the old version to the new one without canceling or restarting them, while ensuring full compliance with current corporate policies at the time of execution. 

System-wide scaling of AI agents becomes the foundation for long-term operational success, enabling progress from a single agent to fully orchestrated, scalable AI systems.

Would you like to learn how to integrate AI agents into your business processes and set up reliable orchestration?

Contact the Integrity Vision team—our specialists will help you find the best solutions for automating and optimizing your business.

Email us at: info@integrity.com.ua