💡What is multi-agent orchestration (MAO)

Because the entire system architecture is modular, the Application layer is completely open. As long as the MAO protocol is met, access can be completed. We plan to implement multiple ends so that MAO can serve different groups of people and different terminals, so that users can access MAO in a variety of scenarios.

MAO: Multi-agent orchestration

MAO is short for Multi-agent orchestration. MAO receives the user's input request, perceives the user's intention, selects the appropriate Agents to collaborate with each other, and finally executes the user's needs and completes the user's task.Intention perceptionWhen the user enters a request, the system needs to perceive the user in order to clearly understand the user's intention, and match the appropriate AI agents according to the user's intention to process the user's request and complete the task.

Orchestrator

The orchestrator is responsible for matching the appropriate AI agent, making orchestration tasks, handing them over to the task manager, and then processing the results of each AI agent's task processing, so as to make decisions for the next step of processing. It is also one of the core components of the entire system and plays a very important role.

Task manager

Task management is responsible for managing the life cycle of all tasks. The Orchestrator can send task management commands to the task manager, such as creating, deleting, and updating. Each task has a state machine to maintain the state. The task manager is one of the core components of the entire system.

Grounding

Grounding is considered the key to artificial intelligence systems achieving true understanding and intelligence. If a system cannot associate symbols with the real world, then it is only manipulating abstract symbols and cannot truly understand the meaning represented by these symbols.

Memory

Memory is to allow LLM to have short-term and long-term memory, which is a very important part in the field of artificial intelligence. Memory can make the system understand the user more intelligently, and can make the user experience, system effect, and performance very good.

Questflow Deagent Protocol

Deagent Guardian validator

Deagent validator is the core component, which is responsible for two responsibilities: receiving external requests, calling agents, and validating agent data.Currently, the Validator serves as the unified entrance for all agent requests, is responsible for processing the gateway of all request processing, and is also responsible for verifying the return data of the selected node agent. After successful verification, the incentive data is calculated. In the future, validators will be more decentralized and introduce more verification nodes to make the entire network more secure, reliable, and transparent.

Deagent Guardian node

Deagent miner node is an instance running an agent, which is the entity that actually provides agent services to the node network. After receiving the request from the verification node, the miner will immediately execute the agent and return the data to the verification node, waiting to be verified.

Deagent Agent registry

Deagent agent registry is the registration management center for all agents. Indicates how many agents are currently providing services in the network. Anyone can apply to become a creator to provide agent services to the network. Once voted on, the agent can be registered with the agent registry to receive incentives.