The video demonstrates Hermes Agent Desktop’s new bot mode, where multiple specialized AI agents collaborate as a team—led by an orchestrator bot and supported by a Bot HR agent that automates team creation—to develop a racing car game from scratch. This multi-agent system enables efficient project planning, task coordination, and parallel expertise contribution, showcasing a promising approach to AI-assisted software development workflows.
The video showcases the new Hermes Agent Desktop bot mode, a feature that enables multiple AI agent profiles to collaborate on projects in a team-like environment. The presenter demonstrates this by building a simple racing car game from scratch using a team of five specialized agents: one orchestrator (Boss Bot) and four worker bots focusing on different aspects such as gameplay, technical art, QA, and WebGL implementation. This multi-agent collaboration is facilitated through a dedicated bot mode interface, which allows agents to communicate via an “agent inbox,” share tasks, and coordinate their work seamlessly.
The presenter explains how to create and customize bots within the Hermes Agent Desktop app, including setting profile pictures, descriptions, and selecting appropriate AI models for each bot based on their roles. A key innovation highlighted is the introduction of a “Bot HR” agent, designed to automatically generate and assign specialized bots for a project based on its requirements. This bot HR streamlines the workflow by intelligently dividing the project into phases and roles, selecting suitable AI models for each task, and fostering efficient team formation without manual setup for every new project.
Once the team is formed, the bots engage in planning and coordination, discussing project phases, dependencies, and technical details such as texture formats and checkpoints. The Boss Bot acts as the project leader, orchestrating the workflow, assigning tasks, and checking in periodically to monitor progress and resolve blockers. The agents work collaboratively, updating documentation, debating plans, and gradually building the game components, demonstrating a new paradigm for AI-assisted project development where specialized agents contribute their expertise in parallel.
Throughout the development process, the presenter observes the bots’ interactions and progress, noting that while the agents do not always explicitly show tool calls, they effectively communicate and advance the project. The team produces early test images and game mechanics, culminating in a functional, playable racing car game with smooth driving controls. Although the graphics remain basic due to the lack of external assets, the project serves as a proof of concept for the potential of Hermes Agent’s bot mode to handle complex, multi-agent workflows in software development.
In conclusion, the video highlights the transformative potential of Hermes Agent Desktop’s bot mode for collaborative AI workflows. By enabling specialized agents to work together under a project leader bot and automating team creation with a bot HR, this mode offers a new way to manage and execute projects efficiently. The presenter acknowledges that the feature is still new and evolving, encouraging viewers to experiment and share tips, while expressing optimism about how bot mode could change the future of AI-assisted project development.