Celonis together with Rollio AI Agents improves process collaboration, and speeds decision-making at legendary spirits maker Campari

https://delivery-p141552-e1488202.adobeaemcloud.com/adobe/assets/urn:aaid:aem:eee3ab05-2232-444f-9b80-58766d95ba44/as/Dise_o_sin_t_tulo__66_.png

At Celosphere 2024, Celonis launched AgentC—a suite of AI agent tools, integrations, and partnerships that enables our community to develop AI agents in the leading AI agent platforms. It also allows them to use AI agents pre-built by partners like Rollio. All these AI agents are powered by Celonis Process Intelligence, making them understand how the business runs and how to make it run better.

Celonis Process Collaboration Agent, powered by Rollio, improves collaborative decision-making by providing a way for teams to engage effectively and work from a single and shared source of truth. It does this by proactively resolving process exceptions through agent-assisted, inter-department collaboration, using a natural language interface. Legendary Italian spirits company Campari will use the Process Collaboration Agent to speed up the removal of credit blocks from sales orders.

“Our partnership with Rollio demonstrates how Celonis enables innovative software companies to build AI agents fed with process intelligence,” said Eugenio Cassiano, SVP strategy and innovation at Celonis. “The Process Collaboration Agent brings together all the relevant people to approve a decision and the process insights to explain the reasoning behind it. By bringing teams together within the collaboration tools that they use on a regular basis, it democratizes the power of process intelligence. We’re excited that this new agent will not only help the Campari Group run their processes more effectively now, but change the way Campari works in the future.”

“The Celonis Process Collaboration Agent is not just a tool; it’s a game-changer for process collaboration,” said Markus Demirci, CEO of Rollio AI. "By harnessing the power of AI Agents together with Large Action Models, we are bringing a new level of intelligence and autonomy to process collaboration. This launch marks a significant milestone in our mission to revolutionize the way businesses handle complex process exceptions today.”

Process exceptions are signals that need action to improve efficiency. Low-complexity exceptions can be managed with simple rules or automation. High-complexity exceptions, however, affect multiple processes and departments, require retrieving lots of information, sharing context among people, and inefficient communication can often lead to lost information across multiple channels. This is the exact problem the Process Collaboration Agent solves.

Consider a complex process like removing credit blocks. Effectively handling credit blocks is a critical aspect of Finance operations. Unnecessary blocks or an inefficient clearance process can lead to delivery delays, canceled orders, delayed payments, and reduced customer satisfaction. Manually clearing blocks is a time-consuming and tedious process—involving many stakeholders in different teams, information scattered across disconnected systems, and multiple communication channels. This leads to increased resolution time, risk of late delivery, increased labor cost, customer dissatisfaction and no auditable log of why the decisions to remove or keep the black in place was made.

Celonis Process Collaboration Agent, powered by Rollio, optimizes the process by providing:

Agent-initiated collaboration: Once a process-exception is identified in Celonis, the agent automatically sets up a “collaboration room” with all involved parties, eliminating the need for humans to start the collaboration.
Process context in natural language: The entire process context is automatically made available by the AI agent in a human-readable fashion. This context is then presented to stakeholders within their enterprise collaboration tool (e.g., Microsoft Teams, Slack, etc.).
Agent-guided decision making: The agent guides and moderates the conversation and upon request makes additional context available. When needed, it also follows up with participants.
Seamless integration: Once a decision is taken, the agent summarizes the conversation and records this along with the decision in the underlying systems.

https://delivery-p141552-e1488202.adobeaemcloud.com/adobe/assets/urn:aaid:aem:578c8558-8725-4056-8819-a17e5436abb0/as/mq3ZMEkAgTQbJdP1QJYFDV.jpg
https://videos.celonis.com/watch/mq3ZMEkAgTQbJdP1QJYFDV

In the real world, the process would look something like this. Rory, a Microsoft Teams Admin, configures the agent directly from within his company’s Microsoft Teams environment. He doesn’t need access to the Rollio backend or to set up any additional connectivity or firewalls. Rory chooses the Celonis data model that contains customer specific business rules, data and logic. He then specifies the associated trigger, which uses logic to identify an alert about significant credit blocks.

https://delivery-p141552-e1488202.adobeaemcloud.com/adobe/assets/urn:aaid:aem:7f9493aa-a032-4650-867f-a650b12f23d8/as/celonis_process_collaboration_agent_01.png

Now, when a block needs to be resolved, the agent starts the process directly within Microsoft Teams. Adele, a credit manager, is automatically added to a group chat with her colleague Lynn, who is a customer service representative, and Nesta, a key account manager. Adele has been automatically identified as the final approver, and she can see at a glance the relevant stakeholders and the reason for the collaboration.

https://delivery-p141552-e1488202.adobeaemcloud.com/adobe/assets/urn:aaid:aem:5fcdc6c8-3445-41a0-a332-495fb230d44d/as/celonis_process_collaboration_agent_02.png

The agent provides relevant customer information (e.g., customer name, order number, order value, credit limit, and requested delivery date) directly within the collaboration. If Adele wants to know more, she can ask the Celonis Process Collaboration Agent about total past-due Account Receivable directly from within the chat window. The agent replies, telling Adele and the others that the customer is close to, but has not exceeded their credit limit.

https://delivery-p141552-e1488202.adobeaemcloud.com/adobe/assets/urn:aaid:aem:64a69a59-8258-4c61-86c7-46fb6c641c12/as/celonis_process_collaboration_agent_03.png

Wanting to investigate further Adele asks Nesta to shed light on the customer’s situation. Nesta responds, saying the customer is going through an ERP upgrade and Accounts Payable is taking longer than normal to process payments. But, she doesn’t feel the customer is at risk of not paying. Adele requests a process to pay. By asking the Celonis Process Collaboration Agent to follow up with Lynn, a separate task is created that Lynn can act on.

https://delivery-p141552-e1488202.adobeaemcloud.com/adobe/assets/urn:aaid:aem:91fc8ca3-ccf4-417d-8368-84c33265d983/as/celonis_process_collaboration_agent_04.png

The Celonis Process Collaboration Agent creates a private chat with Lynn, to follow up on the task of collecting a promise to pay from the customer to resolve the credit block. Lynn can see the relevant details to complete the task (e.g., the order number, overdue Accounts Receivable amount, the requested delivery date, and the invoice number) directly within the Microsoft Teams chat. Once Lynn obtains the promise to pay, she can upload the PDF and notify the Celonis Process Collaboration Agent that it has been collected directly within the same chat window. The underlying LLM parses the PDF and retrieves the important information automatically.

https://delivery-p141552-e1488202.adobeaemcloud.com/adobe/assets/urn:aaid:aem:d9335191-d309-419f-9da5-7cbfa6901c23/as/celonis_process_collaboration_agent_05.png

Back in the main group chat for the credit block, the Celonis Process Collaboration Agent alerts everyone that the promise to pay has been confirmed, including the important information of the amount and date. Satisfied with the outcome, Adele removes the credit block and informs the agent that she has done so. The Celonis Process Collaboration Agent summarizes the reasons for the approval and archives it for future audits.

This is just one example of how Celonis is unlocking the true potential of enterprise AI with AgentC and our collaboration with partners like Rollio.

“In an era where generative AI is reshaping industries, the new Celonis Process Collaboration Agent stands out by moving beyond traditional AI capabilities,” said Demirci. “While previous AI systems primarily served as passive tools for generating text or analyzing data, the Process Collaboration Agent is an active participant in business processes. Leveraging the concept of Collaboration and Large Action Models (LAMs), the Process Collaboration Agent integrates the linguistic fluency of large language models with the capability to perform complex tasks autonomously, thereby transforming the collaborative handling of process exceptions.”

“Celonis’ robust ecosystem and partners like Rollio drive huge benefits for us,” said Laura Buseghin, Process Optimization and Automation Director at Campari Group. “The Process Collaboration Agent will let us be smarter and faster in managing process exceptions, enabling our people to quickly and consistently access information that they would otherwise have to search for across different systems in a complex, time-consuming manner. We’re excited to work with the agent and to assess how we can grow its use across the business. We are always looking for new ideas and new ways to innovate. And we are just getting started. The Celonis Process Collaboration Agent will not only speed up collaboration, but allow us to handle a higher volume of credit blocks as the company grows. When processes work, we can toast to life together.”

To learn more about how Celonis enables companies to create effective AI agents and orchestrate them once they’re up and running, read Manuel Haug’s latest article, How Process Intelligence helps companies build effective AI agents.