In the import-export world, foreign currency exchange risk is very unpredictable due to market volatility. To eliminate the risk, people often buy financial products that allow them to sell foreign currency at a defined target exchange rate.

To solve this problem, we developed an AI solution: Aforex, a web app that integrates information from foreign currency orders, invoices, and hedging financial products, suggesting the best ones to buy.

Results

+1% economic gain generated by the hedges suggested by the algorithm (VS the hedges actually made)
60/65% accuracy of medium-term forecasting algorithms
80/85% accuracy of long-term forecasting algorithms

Getting to know your customers closely is the first step in building an effective sales strategy. AI tools exist that can raise the level of customer knowledge, enhancing promotion and sales activities.

A recommendation engine suggests the “next best product” and the most suitable discounts for specific customer groups, based on purchasing behavior recorded over time (number of pieces/year, spend/year, average discount, etc.).

Results

• +3% ​Marginality per customer

For public lighting facility management, being able to intervene early on the state of deterioration of a support is a priority. In this project, we developed AI models that predict the health status of the supports and the remaining “life”.

By integrating endogenous data with exogenous data (distance from the sea, rainfall, landslide risk, etc.), the AI models calculate the risk of deterioration, prioritizing the interventions to be implemented.

Results

– 40% Risk level in the simulation
– 25% Average budget allocated per municipality

Predicting call center calls is the key to dramatically reduce queues and maximize the user’s experience. This means planning shifts, in the short term, and organizing teams, internal or outsourced, in the long term.

With the Ahead platform, it is possible to predict incoming call flow and organize operational resources based on demand trends, within a single dashboard.

Results

– 37% Average reduction of forecasting error across all levels of detail (daily, hourly, and monthly)