Celonis Games 2025: AI Lab Hackathon

In the opening keynote of Celosphere 2025, Co-founder and Co-CEO Alex Rinke set out Celonis’ vision: that the future of the enterprise is both AI-driven and composable. An enterprise, as he put it “...that can improve continuously – using AI to automate and enhance what you do, every single day.” It was appropriate, therefore, that the event approached its curtain call with the AI Lab Hackathon awards – a real-world demonstration of this vision in action.

Part of the ‘Celonis Games’, the Hackathon finale was an inspiring combination of competition and innovation. Trailblazing teams went head-to-head, showcasing viable prototype solutions that use Celonis process insights and Enterprise AI capabilities to create tangible business benefits. Fully functional solutions designed and built from scratch in just eight hours.

Judged by an expert panel, a pharmaceutical manufacturer and technical supplies sourcing platform Conrad Electronic edged ahead of multiple rival solutions to reach the final round. But while teams had eight hours for the build, they had just three minutes to demonstrate their prototypes’ value to a main-stage audience who would crown the AI Lab Hackathon winner.

Here’s a snapshot of each finalists’ entry:

Pharmaceuticals manufacturer – Agentified E2E Vendor Onboarding

The team from a leading global maker of high-quality pharmaceuticals addressed a significant pain point common to many organizations (particularly manufacturers) – the vendor onboarding process.

The Opportunity

The team highlighted that the onboarding workflow, while vital to smooth operations, is commonly a long, manual, and cumbersome procedure with frequent delays. Vendor onboarding requires coordination across several departments, including Procurement, Legal, Compliance, and Master data and, in the worst cases, can be so time consuming that it doesn’t get completed properly at all. This increases risks of supply chain disruption, higher costs, product non-compliance, and operational inefficiency.

Team’s solution

The team demonstrated the usage of Enterprise AI solutions, powered by Celonis capabilities to overcome these complexities – providing an intelligent end-to-end vendor onboarding process. The core of their solution was a ‘request orchestrator’, designated as the new home for the company's entire vendor onboarding process. This orchestrator’s function was to consolidate, manage, and synchronize all vendor onboarding steps across AI agents, humans from different departments, and multiple systems.

The team consulted with category managers to understand the specific help they needed when reviewing supplier requests. Based on this input, they developed specialized Microsoft AI agents tasked with handling initial activities such as supplier categorization and checking for alternatives. The team described how, when a category manager reviews a new onboarding request, the solution emails them a standardized form summarizing all the necessary information, including AI agents’ assessment and categorization of the request.

This summary equips the category manager to make a swift, data-driven decision to either approve, reject, or move forward with the process. Crucially, the entire automated process was fed back into the Celonis Platform to be monitored and audited. This data acts as a foundation for continuous learning to identify which steps should be further improved or automated, ensuring the solution evolves and maximizes efficiency.

While the team’s Hackathon prototype is in its early infancy, the company believes it can help shape broader AI-driven process optimization initiatives.

Conrad Electronics – Enabling the Intelligent Sourcing Platform

Conrad Electronic is Europe's largest B2B sourcing platform for technical and business supplies. Their team’s solution was designed to optimize supply chain operations.

The opportunity

One of the major challenges in supply chain transformation is disconnected processes. Supply chains are a mega-process made up of many smaller processes in Procurement, Inventory Management, and Order Management. The Conrad team explained the importance of getting the fundamentals right to ensuring optimal operations:

Storing products in the most effective and profitable manner.
Knowing which products in the catalog are ‘hidden champions’ (unsung bestsellers) and which are low performers to be excluded from future purchasing.
Knowing how much (of which) inventory they should buy and at what point in time they should buy it.

While the principles seem simple, executing them in practice, in real-time, at scale across multiple sites, teams, suppliers, and systems is not easy.

Team Conrad’s solution

Conrad’s solution combined traditional process mining techniques with advanced AI to connect the data points between Procurement, Inventory Management, and Order Management. The prototype provides a full end-to-end perspective to inform and empower an intelligent Inventory Management strategy.

The methodology used by Conrad’s new solution is structured into four distinct stages:

Explain: A machine learning model is employed to uncover and explain the core drivers of product profitability.
Analyze: Process mining is used to rigorously analyze the real-time, end-to-end Inventory Management process.
Enrich: AI enriches the data with annotations, enabling the system to provide a clear recommendation on the optimal action required for each specific product.
Act: The system translates these insights into direct executions across:
- Liquidation / cease buying a product
- Increase inventory and marketing
- Continuous monitoring and maintenance of current inventory levels
- Pricing review and carrier cost renegotiation
- Shift procurement and demand to alternative items.

What Conrad’s team built was an Inventory Management control tower that delivered crucial visibility focused on both product profitability and inventory performance. With AI-powered agility, product strategists can swiftly identify and address critical stock issues before they impact commercial performance. As a result this prototype can positively impact working capital, operational efficiency, supplier performance, and customer experience.

And the winner was…Conrad Electronic

The immediate value that can be generated by both Conrad and the pharmaceutical company’s solutions made it evident to everyone watching the Celosphere Games that both solutions have tremendous potential. They demonstrated how companies can use Celonis to analyze, design, and operate AI-driven processes, delivering real value today – not just potential benefits down the road.

After both teams presented their solutions, the winner was decided by way of a ‘highly scientific’ audience volume clapometer – so, huge congratulations to the team from Conrad Electronic, who emerged triumphant from Celosphere 2025’s AI Lab Hackathon final.

Celonis AI Lab Hackathons go global

Celonis holds AI Lab events in cities around the world on a regular basis, not just during Celosphere. Check out the following articles to learn more about how companies are using the Celonis Platform to build Enterprise AI solutions: