Configure Numerical-Analysis AI Agent

I need to get an AI agent up and running whose sole focus is numerical-data analysis. Once it is live, I want to be able to drop a dataset into a folder (or point to a database/API), issue a prompt, and have the agent return clear insights—summary statistics, trends, maybe even an alert if something looks off.

To get there, I will need you to:

• Install and configure a suitable agent framework (LangChain, AutoGPT, CrewAI, or whichever stack you recommend) on my preferred cloud instance.
• Wire it to my data source—initially CSV files, later a SQL endpoint—so the agent can access fresh numerical data without manual uploads.
• Build a reusable analysis workflow that:
– reads the data,
– performs descriptive and trend analysis,
– outputs results in text plus at least one visual (matplotlib, Plotly, or similar).
• Document the whole process in a concise README so I can reproduce or extend the setup.

Acceptance criteria
1. A single command spins up the agent.
2. A sample dataset I provide is processed end-to-end.
3. The agent returns a text summary and a chart, with its reasoning steps logged.

Let me know which libraries you intend to use, how long you’d need, and links to any similar numerical-data agents you’ve delivered before.