Skip to main content

New Version 2.4 - Copilot Multi-Agent Command

· 4 min read
Tiven Wang

We are excited to announce that the latest version of the Metad Analysis Platform, version 2.4, is now available! As a cloud-based agile AI data analysis platform, it integrates multidimensional modeling, indicators management, and BI visualization to provide users with a one-stop data analysis solution. This update introduces the revolutionary "Copilot Multi-Agent Command" feature, further enhancing the platform's intelligence and user experience.

AI Digital Business Experts

In Metad Analysis Platform 2.4, we have introduced a revolutionary feature—the AI Digital Business Experts Platform. This feature aims to digitize professional knowledge and experience from various industries, allowing users to quickly access and apply these valuable resources to build their own enterprise data analysis systems. Whether you are a startup or a large enterprise, this feature enables you to quickly build a data analysis framework tailored to your business needs, improving decision-making efficiency and accuracy.

The platform not only provides users with ready-to-use digital business expert resources but also supports industry experts in digitizing and sharing their expertise with users in need. This sharing and collaboration mechanism promotes the dissemination and application of knowledge, contributing to the progress of various industries.

You can browse more information about AI Digital Business Experts on the AI Digital Business Experts page.

Major Update Features

1. LangGraph Framework

Utilizing the LangGraph framework, we have created stateful, controllable loop processes for Copilot commands with LLM agents. This means a single command execution can call one or multiple LLM agents multiple times. The stateful, controllable loop agents can execute commands in interactive mode, allowing users to intervene in the execution process and control the direction of command execution, ensuring precise and flexible analysis.

2. Dimensional Member Vectorization

We have added dimensional member vectorization in multidimensional models, enabling Copilot command agents to obtain accurate information about dimensional members. This feature enhances the accuracy and efficiency of agents during the data analysis process, ensuring high-quality analysis results for users.

3. Fewshot Similar Example Retrieval

At multiple nodes of the main command agent, we have added the ability to retrieve similar examples using fewshot learning. By switching different business roles, the same command can produce different application effects in various business scenarios, meeting users' diverse needs.

4. Copilot Command Quick Dialog

We have added quick dialog buttons for executing commands on the main command's function interface, allowing users to quickly locate and use the corresponding commands to perform required functions, greatly improving operational convenience and efficiency.

Command Quick Dialog
Copilot Command Quick Window

New Major Commands

1. Modeler

Based on user prompts and table information, and referencing business role examples, the /modeler command can create shared dimensions and multidimensional datasets. This command makes the data modeling process more automated and efficient, allowing users to focus on data analysis and decision-making.

Modeler Command
Modeler Command Structure

2. Indicator System Architect

Based on user prompts and model information, and referencing business role examples, the /indicator-architect command can build an analysis indicators system and create indicators one by one. Users can easily define and manage key business indicators, achieving comprehensive monitoring and analysis of business performance.

Indicator Architect Command
Indicator System Architect

3. Calculation

Based on user prompts and model information, and referencing business role examples, the calculation command can create different types of calculations to meet the needs of story dashboard displays. This command helps users quickly generate the required calculation results, supporting the decision-making process.

Calculation Command
Calculation Command Structure

Command Collection

For the currently available commands in the Metad Analysis Platform, please refer to: https://iwk3znqmad.feishu.cn/slides/X27ds9ZK4l0kqDdRhKgcFVMcnce

Conclusion

The release 2.4 of Metad Analysis Platform marks a new level of intelligence for data analysis platforms. By introducing Copilot Multi-Agent Commands, we provide users with more powerful tools and a more flexible operational experience. Whether you are a novice or an expert in data analysis, you can easily build and manage your data analysis system with this platform. Come experience Metad Analysis Platform 2.4 and discover the charm of intelligent data analysis!