dangbot is an experimental autonomous agent platform.
dangbot is a laboratory for experimenting with new kinds of autonomous agents. It is a place to explore new kinds of organizations by rapidly iterating not on the foundational LLMs at the core, but on the structure within which they are nested and the ways in which many agents can be combined to make teams effective for a variety of purposes.
dangbot is a place to build unique characters and explore their relationships with each other. Characters can be built from templates, or from scratch.
Many Characters can be included in a single chat. This allows Characters to collaborate on projects. Compose the team you need, then get to work.
Set a goal for your team and let dangbot split it into tasks that can be assigned to individual Characters.
Plans can be assigned to individual Characters, or to the whole team. Characters can execute the plan by working together with each other or with humans.
Customize your prompts to make them unique to your characters.
Persistant, transparent, editable memory for each character.
Flexible API integrations for integrating web search, private APIs and more.
Use OpenJourney or Stable Diffusion from the studio to generate character portraits.
Characters, Sources, Templates, Services, and Studios can be shared with other users through the package manager.
Autonomous Agents are still in their infancy. They're not ready for the real world yet, but if we're going to learn how to fold them into our lives effectively, we need to start experimenting with them now.
We're going to need a way for a lot of us, from all walks of life, to experiment with them, and to share our findings with each other more easily than Twitter threads and screenshots.
My hope is that this platform will allow laypeople, artists, and researchers to experiment with Autonomous Agents without having to worry about the technical details of spinning up a server, or implementing a chatbot from scratch. No technical expertise required.