Jason Wang

Thanks to Activant Capital for including Datost (YC P26) in their Cognitive BI market map. Their research is worth reading if you work with data at any level. Some of the findings that validated our thesis at Datost Employees lose 80 hours per year to underperforming IT and 180 hours per year to slow legacy software. Active BI usage across enterprises sits below 26%. The tools get purchased. They collect dust. Text-to-SQL accuracy on academic benchmarks hits 86-94%. On real enterprise databases? 54-63%. On open-ended business questions with raw SQL? As low as 16.7%. This isn't because the models aren't good enough. It's because they don't have enough context. It's an architecture problem, not a model problem. Adding a semantic layer can hit close to 100% accuracy when paired with a governed semantic layer, compared to 16% with raw SQL. A semantic layer that enforces the math. A context l

Thanks to Activant Capital for including Datost (YC P26) in their Cognitive BI market map. Their research is worth reading if you work with data at any level. Some of the findings that validated our thesis at Datost Employees lose 80 hours per year to underperforming IT and 180...