Bridging Data and Business Logic
Leah Weiss, co-founder of Preql, on delivering clean data to LLMs
This mismatch is why modern data projects so often stall between expensive infrastructure and actual decisions. The technical team is trained to manage pipelines, schemas, and code, while the operator or finance lead is trying to answer plain business questions like why margin fell or which product line is working. Preql is built around that handoff problem, using no code workflows and AI agents to turn messy source systems and spreadsheet logic into shared metric definitions that business users can trust and AI tools can query.
-
Earlier semantic layer tools often pushed the hardest work onto analytics engineers, asking them to encode business logic in more code. That failed in many companies because the people closest to metric definitions were finance and operations teams, not the engineers maintaining dbt models and data pipelines.
-
The practical symptom is that real business logic lives in Excel, not in the warehouse. Teams reconcile revenue, headcount, or margin in offline spreadsheets, then struggle to make AI search or chat tools useful because those apps assume the underlying definitions are already clean and consistent.
-
That is also the wedge against tools like Glean and Hebbia. Those products help employees search and synthesize company information, but they sit higher in the stack. Preql is trying to fix the layer underneath, where metrics are defined and cleaned, so the answer engine does not return polished nonsense.
The next phase of the market moves from dashboards for humans to decision systems and copilots that can act. That raises the value of products that capture business definitions in a form machines can use. The winners will be the platforms that let finance and operations teams express their logic directly, without waiting on a specialized data team translation layer.