Exports of agronomic products are a major source of income for many countries across the world. Import, export, and domestic usage data, as well as the adjustments to production and marketing that follow, may all be better predicted using monthly Agronomic Export Forecasting. To better anticipate the growth and drop of Agronomic exportations, this study presents a new approach called Agronomic exports time series -longshortterm memory. An algorithm is used to train vectors of words by dividing words into groups and then using Term Frequency -Inverse Document Frequency /word cloud to study informational keywords. This study investigates whether the AETS -LSTM model can effectively use the purchasing managers' index (PMI) of every industries to anticipate the increase and fall of agronomic exports. A study of the PMI principles in the financial and insurance industries found that using keyword vectors increased the accuracy of predicting growth and decline in Agronomic exports by 82.61 %. Combining electrical and optical keywords improves its effectiveness in these categories. Thus, agribusiness operators and policymakers will be able to use the recommended approach for a more accurate assessment of local and international output and sales.