Aquaculture runs on timing. Knowing what demand will look like weeks ahead shapes every decision, from feed orders to harvest plans to pricing. For a long time that knowledge lived in experience and phone calls. We wanted to see how much of it a model could learn.
What the model looks at
Shrimp AI was custom-trained for one job: predicting demand. To do it, the model scans the country using three kinds of signal that tell a story when you read them together.
- Weather: the conditions that shape pond activity from region to region.
- Pond colour: a visible clue to what is happening in the water.
- Pond shape: how ponds are laid out and used across the landscape.
None of these signals is new on its own. What is new is reading all of them, everywhere, consistently, and turning them into a forecast a business can act on.
Why custom-trained matters
General-purpose AI is impressive, but it does not know what a pond should look like in the middle of a season. A model trained for this specific problem learns the patterns that matter to the industry and ignores the noise that does not. That focus is what makes its output useful rather than merely interesting.
Read the landscape, predict the market
Shrimp AI turns weather, pond colour and pond shape into demand predictions, giving aquaculture businesses a view of the season before it arrives.
From prediction to decision
A forecast is only valuable if it changes what people do. We designed Shrimp AI so its predictions feed straight into planning conversations: where demand is likely to rise, where it may soften, and where a business should pay closer attention.

“The best AI projects do not replace expertise. They give experienced people a wider view than any one person could have.”
— Softexedge Team
What it means for other industries
Shrimp AI is built for aquaculture, but the approach travels well. Many industries have signals hiding in plain sight, in images, weather and everyday operations. If an important decision depends on reading many small clues at once, a focused model may be able to help.
If you are curious what that could look like for your business, we would love to talk it through.










