NeoSapiens was inspired by the idea of combining AI and hierarchical swarm intelligence to automate and orchestrate the problem-solving process, potentially benefiting a large number of people.
NeoSapiens automates the creation of domain-expert agents needed to solve a given problem. The problem is broken down into smaller steps, and each agent solves the steps related to its area of expertise. The outputs are then aggregated to provide a coherent solution.
We built NeoSapiens using open-source AI tools, cutting-edge machine learning research, and custom swarm intelligence logic. The team integrated an open-source swarm orchestration framework to enable agent communication and coordination.
Challenges included ensuring seamless collaboration between autonomous agents and optimizing AI models for real-time problem-solving while balancing efficiency and effectiveness.
We are proud of the tools we equipped our swarm with to solve problems through common developer tools like terminal and browser. Watching intelligent agents collaborate to find solutions has been rewarding, and we believe our project can make a real impact.
Throughout the journey, we learned about working with Large Language Models, swarm intelligence, and the importance and challenges of teamwork and coordination in problem-solving. We also learned how to overcome technical hurdles and create effective strategies to tackle a broad array of problems.
As we move forward, we have several exciting plans to enhance NeoSapiens. We will fine-tune our AI models to make the agents smarter and more effective. Creating a database of tools will allow agents to retrieve the most relevant tool for each problem. Improving the user interface with voice commands will make the system more user-friendly. Automating tool creation at runtime will increase efficiency, while expanding the logic to have synchronously running agents when appropriate will enable tackling more complex problems. Additionally, we are working on implementing a long-term memory shared protocol, better function calling, and improvements to the boss agent for more effective worker agent creation. These enhancements will not only allow us to address an even wider range of challenges but also make NeoSapiens a more powerful and versatile problem-solving platform.
Log in or sign up for Devpost to join the conversation.