Creating an intelligent reality with Cogine
Reality World is a multi-agent platform built with interoperability as its core, in which a dynamic agent or program can be added and interoperated with all other agents. It's like a complex system and emerges digital intelligence.
Reality World Open Source Conference
Computing Innovation
Conversational
Code can talk to user, system or other agents to ask more information at any function location, the agent will be paused and continue to run when got response.
Interoperability
Agents written by different developers could be running in a same memory environment and interoperate with each other with the permission of user.
Self-evolving
The agents are working in a collaborated way and user could custimize it's personal composation to make the whole system self-evolving and emerge digital intelligence.
Logic-abstraction
We re-organized the whole program to make developers focused on logic of busniess and forget about the hardware, data and programing language abstraction.
Let everyone can write AI agent
Programming is not easy, but the hardest part is not the programming language itself, its the program structure organizing. We build a high level logic-abstraction to hide the structure and let developer focused on logic only.
Most visual graph editting is just a visual wrapper of the original codes. We build a graph interaction which is as simple as Houdini and as powerful as any turning completed programming language.
An agent VM, platform and simulator
We defined a agent format standard to represent a turing completed agent program which can be loaded and runned dynamically so everyone can develop an agent to handle personal needs.
We then build a platform with a virtual machine to interpret and execute the above agent format program dynamically. By this dynamical capability, developer can create and share agent with users as simple as share a url.
Multi-agent conversational collaborating framework
Agents can commnuicating with user, system and other agents in a very simple way which provides the ability to finish a task in a collaberating way.
In a LLM-powered application, the inputs of a function is semantic instead of structured data, which brings uncertainty and needs conversational way to get more information. For solve this problem, we propose a new concept and a novel architecture called "Conversational computing" for the first time in the world, to make any function can be paused and conversational commnuicating with user, system and other agents to get full information it needs without extra works.
An engine of complexity to emerge intelligence
In biology or social economic syetm, many agreements have come to that the intelligence of the system come from the emergence mechanism of the complex system theory.
In computer science and engineering, applications are isolated each other for data security and some other business reasons. The lack of interoperability make the applications non-cooperative and hence none-intelligent. We re-organized the structure and execution of software contruction process to provide a interopability-native architecture to form an engine of complexity.
