AIgora

Our project supports investors and business owners in making informed decisions through a multi-agent, two-step process.

First, specialized agents perform parallel analyses: one evaluates financial aspects, while the other examines societal and market impacts. Next, an aggregation agent synthesizes these findings. Then, two additional agents assess pros and cons, enabling quick and actionable insights. A final agent consolidates the positive and negative arguments to provide the user with the most relevant recommendations.

By parallelizing tasks wherever possible, the system delivers efficient and comprehensive decision support.

Currently, our agents focus on financial and societal aspects. In future works, we can add an environmental agent and others to broaden the scope and provide even more comprehensive analyses.

Initially, our frontend displayed each step of the thinking process: analysis agents, pros and cons agents, and synthesizer agents. It provided users with insights into the procedure, including a loading bar. However, this approach did not fully leverage the potential of parallel and sequential agents, as we were calling them one by one. This created a difficult trade-off between speed and user experience.

Since code quality and structure were key judging criteria, we aimed to make agent interactions as efficient as possible. To enhance user experience, we began integrating insights from the agents’ reasoning, paving the way for a multi-agent system that effectively supports investors and business owners.

Due to integrations issues we had to change the frontend last minute making it less stylish than we hoped it would be.