Adequate Data turns a plain-English question into a clear answer: what the result is, what data was used, which filters applied, and how the calculation was made.
The product still supports natural language questions, but the marketable promise is verification: supported answers should show their work.
Adequate Data is designed to make answers inspectable, reproducible, and auditable rather than claiming that AI output is automatically correct.
Surface the source table or file, selected fields, row counts where available, and data freshness context.
Keep filters, calculations, assumptions, and transformation paths close to the answer so the result can be challenged.
Give the answer a reviewable ID so the proof file points back to the specific work product.
Export the evidence needed to show how a metric explanation was produced and what remains unverified.
Adequate Data should be evaluated against documented controls: authenticated access, encrypted connections where configured, credential encryption at rest, audit logging, and proof exports for supported workflows.
Review the proof trail