Data Tech 2026

Should you use an off-the-shelf AI service? Compose narrow tools into a pipeline? Strip an open-source model for parts? Build a chat--bot?   For one SDG client, each of these has been the correct answer to a different problem. Through four short case studies spanning video, analytics, accounting, and ETL, this talk offers a practical framework for matching AI tools to problems. The core claim: in production AI, the architecture decision usually matters more than the model decision.