Leaders of AI: a multi-agent system with 60 specialists in just one day.
At a glanceStarting point
How do AI employees work not individually but as a team?
Developing working AI teams and multi-agent systems presented several complex challenges: the technology was still very new and unexplored, which made practical use cases hard to validate. There was a lack of understanding of how AI employees can work together not only individually but also as a team. It was also unclear how human ways of thinking, decision logic and collaboration skills can be successfully transferred into AI systems.
The central question: how do AI teams produce genuinely valuable output, how does collaboration between AI employees work, and how do people bring their collaboration skills into AI systems?
Solution
A prototype months in advance, five specialisations in one day.
The idea for the workshop came from an initiative by Leaders of AI, EY (Ernst & Young AG) and Virtual Identity, who wanted to plan an innovative event to explore multi-agent systems together. scoreprise.AI was invited alongside ThynkAI as a technical partner to help build the multi-agent system and later to contribute to the workshop event.
Months before the event, Leaders of AI, EY, ThynkAI and scoreprise.AI jointly developed a working prototype of a multi-agent system to serve as a solid basis for the planned group work. Virtual Identity took on the organisation of the event as a workshop format with EY clients in the Leaders Lab.
On 21 February 2025, a one-day event took place in the EY Wave Space with 60 experts from a wide range of fields: prompt experts, management experts, AI experts and marketing professionals. The 60 participants were divided into five groups, which did not have to start from scratch but could work with the prototype that was already complete.
Each group was given the task of developing this base prototype in a particular direction and giving it an individual specialisation. Under the guidance of Leaders of AI, EY, ThynkAI and scoreprise.AI, five different multi-agent systems were created, each with its own character.
The underlying multi-agent system was designed for a fictitious company and its task was to monitor and understand the market in the area of social media and marketing and to develop corresponding recommendations for content strategies. The hierarchical structure had an AI CEO at the top, who made decisions and delegated tasks to specialised AI employees.
Preparation
A prototype, months in advance
Before the event, Leaders of AI, EY, ThynkAI and scoreprise.AI built a working multi-agent system so that the groups did not have to start from scratch.
Leaders Lab
Five groups, five directions
On 21 February 2025, 60 specialists in five groups each developed the base prototype further into its own specialisation.
Result
AI employees in a team
By the end of the day there were five systems for market monitoring and content strategy. The best solutions were awarded prizes.
The aha moment
Today, working prototypes are created in days, not months.
While many companies associate the introduction of AI with processes lasting months and years, 60 specialists developed five specialised AI teams in one day on a prepared base prototype. The key was not technical expertise but the mix of disciplines.
Result
Five systems, awarded prizes in the evening.
The project led to several important insights: it was successfully demonstrated that AI employees already work with one another today and can collaborate as a team. The basis developed by the partners made it possible for five different groups to develop specialised multi-agent systems in just one day.
The combination of months of preparation and a single day of specialisation exceeded all expectations: at the end of the day, five groups presented their individually shaped multi-agent systems for market monitoring and content strategy development. The best solutions were awarded prizes and demonstrated the range of possible application scenarios.
Well-known EY clients such as Red Bull were able to see the practical applicability for themselves and showed concrete interest in implementing AI departments. The project validated the hypothesis that in future every company will use AI employees and smaller AI teams, with not only people collaborating with AI but also AI employees collaborating with one another.
The successful collaboration between Leaders of AI, EY, Virtual Identity and scoreprise.AI showed how, by combining different areas of expertise, innovative AI solutions can be developed in record time.
More use cases.
All use cases
M&A advisory
KERN Unternehmensnachfolge: invoices, proposals and NDAs from a Teams message.
Higher education and research · Project
Leuphana: How human may a chatbot appear? A study with 500 students.
Media and education · Project
campus a: Journalistic feedback for Austria's first nationwide school newspaper.
Next step
Your use case, shown on a comparable one.
Cancellable monthly · Operated in Germany