AI Innovation comes alive in Madrid: The latest from the Celonis AI Lab

In September, our Madrid office was the stage for a very special event: the AI Lab. On this occasion, seven of our customers joined this series of AI hackathons to work side-by-side with Celonis experts. The goal was clear: to design and prototype innovative solutions using the unique process data and business context provided by Celonis and our latest AI tools, such as Process Copilots, AI Annotation Builder, Orchestration Engine, and Prediction Builder.

After an intense day of work, each team had the opportunity to present their project in just five minutes. They explained the business problem they addressed, the potential impact of their solution, and, most importantly, how they creatively used AI to achieve it. The AI Lab proved to be much more than a technology event; it was a space for collaboration, healthy competition, and co-innovation.

Here is a summary of the solutions that were created:

Leading global credit insurance operator

Challenge: The team wanted to optimize the company’s process for handling blocked invoices. Instead of relying on weekly reports and manual interactions between the Finance department and individual business units, the team would give both groups real-time, end-to-end visibility into the process and increase the automation rate.

Solution: The team deployed an AI assistant (using Process Copilots) to improve communication and visibility. Now, the finance team and the business units can instantly get up-to-date information on the status of any invoice, and if necessary, the Process Copilot can draft an email to explain to the customer why their invoice has not been paid. Furthermore, the AI Annotation Builder automatically analyzes invoices to assign the necessary actions, optimizing the workflow and minimizing manual effort.

Regional Government

Challenge: By regional law, all tax refund decisions, whether a refund is granted or not, must be manually ratified by an official. On average, ratification takes two days. But if this process is delayed, the company could be out of compliance with the legally required deadlines and be subject to interest charges.

Solution: Using Orchestration Engine, the team streamlined the refund process. If a decision is not ratified within two days, a complete summary of the refund case, which includes the actual decision and activity history, is automatically sent to the official. Once they have ratified the decision, the system automatically updates the refund status, ensuring visibility and compliance with established deadlines. Everything adheres to and is confined within the process context to facilitate the monitoring and auditability of tasks, people, and AI agents.

Navantia - International leader in naval construction and defense

Challenge: Component shortages can interrupt shipbuilding. With real-time inventory visibility, Navantia construction and logistics personnel could detect and respond to component availability problems and ensure production continuity.

Solution: Using Celonis Process Intelligence, Process Copilots and Action Flows, the Navania team created a proactive material management solution.A Process Copilot chatbot lets people ask about the status of work orders through a natural language interface. The operational view offers a mass perspective of orders with missing materials, enabling teams to make quick and coordinated decisions. Additionally, automations (Action Flows) instantly notify the involved teams to act immediately.

European luxury goods corporation

Challenge: This company’s accounts payable (AP) department gets thousands of emails each month from suppliers requesting the status of invoices awaiting payment. Traditionally, responding to these emails has been a time-consuming, manual task.

Solution: Using the Machine Learning Workbench and AI Annotation Builder, the team built a solution that extracts key details from each email, automatically identifies the related order number, and assigns a response category. From there, an Action Flow uses AI to draft a customized reply email based on one of several response templates. Finally, an analytical view provides complete context to understand and continuously optimize the process.

European luxury goods corporation

Challenge: The company wanted to reduce the number of orders rejected due to stock-related problems by improving inventory management, particularly between plants in different countries.

Solution: The team proposed a solution that uses a machine learning algorithm and the AI Annotation Builder to optimize the product allocation process. This system directly analyzes 65% of the assignments based on the company's logic: customer prioritization, order value, and available stock. The tool automatically proposes solutions, such as reassigning reserved stock or accelerating quality control, which are communicated via email.

Spanish insurance company

Challenge: As part of its life insurance business, this company covers a variety of funeral service expenses, including headstones. Producing a headstone is a complex, multi-stage process, and a delay at any stage could result in the headstone not being delivered on time and cause undue distress for those already suffering a significant loss.

Solution: Using the AI Annotation Builder, the team developed a solution that collects structured and unstructured data from each order, analyzes the information, and categorizes the order based on predetermined criteria, such as a delay. The solution can then propose a next step. Order managers can review the results in an operational view and, from there, activate an Action Flow to send an email to the person responsible for the delayed stage of the process.

WINNER: Insud Pharma - Multinational pharmaceutical company

Challenge: The Insud Pharma team wanted to improve visibility across the company’s supply chain, specifically for orders at risk of delay and critical orders that were completed and ready to be picked up for delivery.

Solution: The team developed an AI assistant (Process Copilot) that provides the Customer Service team with visibility into order status. In addition, a pilot was conducted with predictive models (built with Prediction Builder) to anticipate orders that could be delayed and proactively identify possible root causes. With this information, the solution will be able to offer corrective action recommendations to notify customers, coordinate teams, and optimize pick-up times, which could eventually translate into reduced storage costs.

Coming Soon to Celonis AI Labs

The solutions don't stop here. At an AI Lab held in our New York City office, an American multinational investment bank won big for their use of our Prediction Builder, addressing know your customer (KYC) case resolutions.

Stay tuned for the 2026 AI Labs calendar, which will feature more customers and partners, more AI, and new innovations.