Guides and field notes
First, what you need before deployment: the sequence, the methodology and the formats. Below, field notes: what we learned while fixing other people's data.
Seven steps; you can stop after any of them
How effect is measured, and when we refuse
What the export must contain for a calculation to mean anything
What we connect to and how data comes in
The terms the system computes in
Read-only access, webhooks, file formats
Field notes
How to De-Anonymize Transformer Behavior on Time Series
A method for interpreting transformer models on time series: Integrated Gradients heatmaps in time × scale × signal coordinates, two verifiable tests, and practical applications for diagnosing data drift.
Retail 2030: Multi-Agent Systems, Regional Intelligence, and the End of the Excel Era
How AI agents transform retail: from demand forecasting to autonomous supply chain management. Territory as a living organism, multi-agent architecture, and why 59% of inventory in a typical chain is frozen capital.
Artificial Intelligence 2026: The Reality of AGI and Engineering Constraints
Why current AI is a plateau, not an exponential. The gap between demos and reality, the reliability problem in long chains, the economics of progress, and what it all means for retail.
Data Flows in Pharma Retail: Modern Challenges and Solutions
Cross-verification of pharma marketing data: reporting lag problems, directory matching tactics, and why automated data flows dramatically change the quality of decisions.
Show us one critical process. We will show how it runs here.
We look at your cycle: how an order is assembled today, who decides, where time leaks and what the system takes over.