Analytics Agent

Current Analytics challenges and its solution

Organizations struggle with slow, fragmented analytics processes that require multiple tools, manual data handling, technical expertise and coordination amongst multiple individuals. This creates barriers to insight generation and adds delay in achieving the goal. The Analytics Agent solves this by filling the skill and access gaps through orchestrated AI agents built with IBM watsonx Orchestrate. It enables users of any skill level to quickly explore data, uncover insights, and make informed decisions, accelerating research, collaboration, and innovation across teams.

Below are the target users for this solution

  • Business Analysts: They will be the most benefited users from this solution as this will remove any technical barrier and help them directly interact with the data from several places. They can even load the data into staging database.
  • Data Scientists and Engineers: This group of people will get benefit from efficiency as they can offload some of their repeatitipve task to Analytics Agent. And this will also speed up data preparation and collaboration.
  • Production Support Teams and Administrators: This solution will help them cover for both skill and efficiency as they can monitor things and answer user queries from single place

The agent acts as a virtual analytics assistant that users talk to, guide, and collaborate with — no coding or complex tool switching required.

Its ability to combine end-to-end automation with human-AI collaboration makes data insights accessible, fast, and actionable for anyone — from non-technical teams to expert analysts — bridging the gap between data complexity and real-world decision-making.

The solution uses a group of AI agents. All the different agents are managed by a manager agent “Analytics Agent”. Below are the different individual agents which serves to the manager agent.

  • File Analytics: Helps to analyze a file and if needed can load the file into staging database.
  • Database Analytics: Helps to get information and insights from the database. Agent creates SQL based on the natural language inputs from the user.
  • BI Analytics: Connects to a BI tool and brings back the requested information.

Analytics Agent understands what a user wants to do and helps the user in getting to its goal. If the user is interested in analyzing a file, user need not to open the file, rather can ask questions and the agent will respond with he required information. Similarly, if someone wants to do some analysis on database, no need to write long SQLs, just ask the question and get the answer in tabular format. If the user wants to get insights from a BI tool, just ask the question and the agent will direct to the right place.

Analytics agent doesn’t replace any of the existing infrastructure, rather it reuses them. The agent is not a completely autonomous agent, it is more like a collaborator or your assistant who can help to move fast while you have complete control.

IBM watsonx Orchestrate:

Our solution, Analytics Agent, harnesses agentic AI through IBM watsonx Orchestrate to revolutionize end-to-end data analytics. Multiple specialized AI agents collaborate under orchestrated workflows to automate tasks including data ingestion, cleaning, analysis, and visualization. Users interact naturally via conversational commands, guiding the agent while it autonomously performs complex analytics processes.

Below is the detailed description of each component of this solution

  1. Analytics Agent: This is a the manger agent. It uses a tool named “Analytics Path” developed using flow builder. This tool interacts with user using user activity form and based on the inputs it understand and routes the request to the correct agent.
  2. File Analytics: This agent is solely focused on the analysis of file, it will return metadata of the file and if you want to go into detail it will help you query the file. It uses two python based tools “File_Reader” and “File_Query”. WatsonX Orchestrate and the LLM engine helps understand the user query and converts them into proper parameters to be passed to these tools.
  3. File Loader: This agent is used to load the file into a staging database
  4. Database Analytics: This agent helps in querying the database across multiple table.
  5. BI Analytics: Helps the user find answers to its question from a report or dashboard. Based on the user’s need it will bring in the precise response or it can guide the user to a report or dashboard.

Code repository
GitHub – RahulGanguli24/WatsonX-Orchestrate-Analytics-Agent: WatsonX Orchestrate Analytics Agent

Team & Members

  • Team Name: Cognos Partners
  • Members: Rahul Ganguli, Vinay Raghuwanshi, Ankit Bhawsar

Appendix

Data files stored in IBM Cloud Storage:

Test Case Evaluation: