jevbooks

← All projects

pg-jev

realZachi/pg-jev · Homepage

Ask Postgres tables plain-language questions with calibrated probabilities.

What it is

A PostgreSQL extension that lets you filter, rank and classify table rows with plain-language conditions, using TypeSafe's Jev model instead of embeddings or vector columns. Aimed at SQL users who want semantic queries over ordinary tables.

How it uses Jev

Jev judges each row as a yes/no Noul question, batched 20 rows per shared state, returning calibrated probabilities. jev() uses them as a boolean WHERE predicate; jev_prob, jev_score and jev_choice expose probabilities, ordered-level positions and most likely options in SQL.

Primitives:noulchoicescore

Technique worth stealing

Batch ~20 rows into one shared state with one question per row to amortise request overhead, and cache answers per row content.

Try it

pgxn install jev, then psql -c "CREATE EXTENSION jev CASCADE" and set jev.api_key.

View on GitHub

judged by Jevjev-1.13.0

Evidence

Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.

  • Jev-centricyes0.87
  • Shows a System One patternno0.19
  • Handles uncertaintyno0.05
  • Measuredno0.06
  • Runnableyes0.98
  • Worth recommendingno0.38
  • Model replicano0.05
  • Problem scopescore on a 0–2 scale1.04
  • About Jevyes0.98

Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.