Trust indicators
- EU business: Cognity SIA, Latvia.
- GDPR-aware privacy and DPA documents.
- Forecasting runs inside Cognity infrastructure.
- AI-generated explanations are optional and can be disabled.
- No raw business data is sent to external AI providers.
Forecasting architecture
Cognity's core forecasting is performed with internally hosted XGBoost models. Forecast training and prediction run inside Cognity infrastructure. Customer sales, inventory, component, supplier, and purchasing data are used to provide the forecasting and planning service, not to train external general-purpose AI models. Cognity's core forecasting and planning functionality does not depend on external AI services.
AI Usage
External AI is used only to turn already computed forecast results into natural-language explanation text. It does not calculate the forecast, train the forecast model, or analyze raw customer datasets.
External AI calls are handled as a separate, controlled component with no direct access to the core system, database, or internal processing logic. Only explicitly prepared, abstracted inputs are provided, and no direct interaction with underlying data sources is possible.
When an explanation is generated, the input is limited to abstracted or aggregated analytical signals such as grouped SHAP or feature contribution summaries: pricing impact, seasonality impact, trend impact, and promotion impact.
We do not send raw sales transactions, supplier lists, customer lists, SKU-level datasets, detailed purchasing plans, or uploaded source files to external AI providers.
External AI providers used for explanations are configured not to use submitted data for model training.
AI-generated explanations can be disabled for organizations with stricter internal data policies.
Data Flow Overview
1. Customer data is uploaded securely
Sales, product, inventory, component, and planning data are uploaded or connected to Cognity.
2. Forecasting and training run internally
Internally managed XGBoost models train and predict within Cognity infrastructure.
3. Forecast outputs and driver signals are generated
Cognity produces forecast results and aggregated driver signals for planning and explanation.
4. Optional AI explanation layer summarizes signals
If enabled, external AI converts aggregated driver signals into readable summaries.
5. No raw datasets are transmitted externally
Raw datasets, SKU-level records, purchasing plans, and uploaded source files stay out of the external AI explanation layer.
Security posture
Cognity applies technical and organizational measures intended to protect customer data against unauthorized access, disclosure, alteration, and loss. Access is limited to people and systems that need it to provide, support, secure, and maintain the platform.
We do not claim certifications we have not completed. Security reviews, subprocessors, and customer-specific data handling questions can be discussed at info@cognityapp.com.