Advancing Multi-Agent AI Frameworks: CBAIA and CAMEL-AI Successfully Host First Flagship Conference of 2025

London, April 26, 2025

The China-Britain Artificial Intelligence Association (CBAIA) successfully held its first major conference of 2025 at the Sammy Ofer Centre, London Business School. Titled “Advancing Multi-Agent AI Framework for Real World Applications,” the event was co-hosted by CBAIA and the China Club of London Business School. The conference brought together professionals and researchers from academia, industry, and the AI developer community to explore the latest in AI agent systems and their real-world applications. As AI agents and multi-agent frameworks gain momentum as one of the most transformative frontiers in artificial intelligence, this conference served as a timely platform for knowledge exchange, collaboration, and innovation. It featured keynote speeches, project showcases, panel discussions, and vibrant networking sessions.

Figure 1: Group photo of the conference participants

A Bridge Between Nations, Ideas, and Innovation

CBAIA is a UK-based organization committed to strengthening academic and industrial collaboration between China and the UK in the field of artificial intelligence. By connecting researchers, developers, entrepreneurs, and investors from both countries, CBAIA promotes knowledge sharing and fosters joint innovation.

The China Club of London Business School, a co-organizer of the event, is one of the institution’s most dynamic and active student societies, with more than 3,000 members. It plays a pivotal role in building cultural and business bridges between China and the global community. This partnership with CBAIA reflects both organizations’ commitment to supporting emerging technologies and international cooperation.

A Highlight: CAMEL-AI Takes Centre Stage

One of the most anticipated highlights of the conference was the participation of Dr. Guohao Li, founder of CAMEL-AI and a leading expert in multi-agent AI systems, along with his core team. CAMEL-AI is the world’s first open-source community dedicated to Large Language Model (LLM)-based multi-agent systems. It focuses on exploring the principles behind large-scale agent collaboration and aims to advance both applied and fundamental research in areas such as task automation, world modelling, and data generation.

Following a brief welcome by Jingwei Zhang of the London Stock Exchange, Dr. Guohao Li delivered the keynote speech titled “AI Agents: The Future of Intelligence”. In his talk, Dr. Li delved into the theoretical foundations and practical potential of AI agents, arguing that multi-agent frameworks are a key pathway to evolving LLMs into collaborative general intelligence systems with intentionality.

Figure 2: Dr. Guohao Li, Founder of CAMEL-AI

Showcasing the CAMEL-AI Ecosystem

One of the highlights of the conference was CAMEL-AI’s live demonstration of its three flagship projects, which together illustrate the diverse applications and flexibility of multi-agent frameworks:

• OWL (Optimized Workforce Learning): An open-source multi-agent collaboration system that mirrors human team-based workflows. By decomposing complex tasks into subtasks and assigning them to specialized agents, OWL enables scalable, automated problem-solving. OWL has become a cornerstone of CAMEL-AI’s approach to task orchestration and has inspired numerous open-source replications and innovations in the AI community.
• EigentBOT: A technical question-answering agent designed for enterprise use. EigentBOT can be instantly deployed to platforms like Slack and Discord and can respond to user queries based on an organization’s own documentation or knowledge base. This tool is particularly valuable for customer support teams, internal help desks, and onboarding processes, where consistent, accurate, and fast information delivery is key.
• Matrix: A social media behaviour simulator powered by CAMEL-AI’s OASIS platform. Matrix replicates the activity patterns of real users on platforms like Twitter (now X), including likes, reposts, comments, and even generating celebrity reactions, humorous memes, or occasional toxic responses. The platform is designed to test social algorithms, model influence dynamics, and simulate engagement strategies for marketing teams and policy researchers alike.

These projects not only demonstrate CAMEL-AI’s technological capabilities but also signal the broader shift in AI research toward agent-based modelling, behavioural realism, and scalable infrastructure for simulation and automation.

Figure 3.  CAMEL-AI team members presenting on site

Figure 4. CAMEL-AI team members presenting on site

Expert Contributions from Industry and Academia

The conference also featured William Lee, CEO and Co-founder of Acl.dev, who introduced their open-source tool infrastructure designed for AI agents. Dr. Xinghui Tao from the University of Cambridge discussed the industrial applications of AI agents in sectors such as manufacturing and energy. Jingwei Zhang of the London Stock Exchange Group showcased how reasoning-based large models are being integrated into financial AI agents, drawing strong interest from the audience.

Figure 5: William Lee (Acl.dev)

Figure 6: Dr. Xinghui Tao (University of Cambridge)

Figure 7: Jingwei Zhang (London Stock Exchange Group)

Panel Discussion: Defining the Next Chapter of AI

During the panel discussion, moderated by Jingwei Zhang (London Stock Exchange Group), key speakers including Dr. Song Hou (Viridien / University of Cambridge), Dr. Guohao Li (CAMEL-AI), Dr. Aran Batth (Moody’s), William (Acl.dev) and Dr. Xinghui Tao (University of Cambridge) exchanged insights on the theme “How Will AI Agents Define the Next Phase of AI Applications?” The panellists agreed that multi-agent systems would play a transformative role in the coming years, particularly in domains such as office automation, intelligent customer service, and social content generation—marking a new chapter for AI deployment.

Figure 8. Panel Discussion

Interactive Networking Session: Shared Ideas, Future Collaborations

In the concluding networking session, participants and speakers engaged in vibrant, in-depth discussions on topics ranging from project collaboration and community engagement to technical implementation details. Attendees posed thought-provoking questions and offered valuable insights, while guest speakers shared real-world experiences and addressed challenges in project execution. The lively atmosphere underscored the collaborative and inclusive spirit of the event and sparked exciting possibilities for cross-disciplinary partnerships. The session emerged as a standout moment of the conference, yielding both intellectual value and potential for future cooperation.