How to analyze thousands of products in stock and generate individual recommendations for each one

I have the following challenge, I am analyzing a csv with several kpis of products such as revenue, volume sold, inventory turnover, inventory coverage and etc.

I need the agent to read this csv and go through each product and create a strategic recommendation for one of the products.

Example: If the product has a high sales volume, a
average sales per day and the stock coverage is low, it indicates that it has a risk of running out of stock and losing sales. Or if we have a lot of stock and very low turnover or the stock is stopped, recommend liquidation.

I tried to process by passing the csv to the agent, I tried to pass in Json, in smaller batches and I always have context problems or the LLM processes only 10 items and closes for no reason.

Is there any way to do this kind of analysis 1 to 1?

I see this as a deterministic rule-based business problem and doesn’t really require an agent to solve it.

One option is to use a Flow Introduction - CrewAI this will enable you to pass each row to a crew in turn. You can then follow up with a review agent to see and report on any issue