VQS

Geometry

Query-free measures over the vectors themselves: cohesion, spread, margin, hubness, redundancy, integrity, and split penalty. This is where chunk-splitting damage and near-duplicate pileups surface, the two failures that quietly cost you the most recall.

Writing quality

Four per-chunk measures, worst one first in your own dashboard: substance (real information, not filler), specific (concrete, not vague), varied (not the same sentence shape on repeat), self-contained (makes sense without the paragraph around it). A chunk that reads like a fragment of a table of contents scores like one, because that is what it will retrieve like.

What to do about a low score

Four moves, all of them yours. Split or rewrite the chunks flagged as incoherent. Drop the near-duplicates so they stop competing with each other. Give every fact that varies by version, date, or tier its own key in the text, so the chunk stands on its own instead of reading as a contradiction. Then add material where the map is thin.

Which layer your problem is in

This is the part worth the read. A low score means better retrieval will not save you: the material is the bottleneck, and fixing it comes first. A high score with answers still going missing is a retrieval problem, and that is where Max-grade retrieval earns its keep. Buying Max to paper over a weak corpus is money lit on fire.

Scored on the real engine

VQS is not a heuristic run over your text in a notebook. Your corpus is embedded by the same engine that answers your queries, on the same owned NVIDIA DGX hardware, so the geometry you are scored on is the geometry retrieval actually sees.

Live in the dashboard

Open Quality. Every corpus you own, side by side, coloured by score, including the ones you have not measured yet. Drill into one and its own 3D atlas shows structure as you built it, quality as colour, contradictions drawn as red arcs between chunks that cannot both be right. Then paste your own examples and see how well they would actually rank, scored live, no production traffic required.

One honest limit

VQS scores whether your corpus is capable of answering well, not whether it covers one specific question, and it's relative, not a certified absolute. A contracts corpus can score high and still say nothing about last quarter's pricing, because pricing was never in it. That's a coverage gap, not a quality problem. Even a great corpus has edges.

Score a corpus

Upload documents, then open Answers and Quality. Turbo embeddings are free and the free tier includes a corpus allowance, so the first score costs nothing.

60 SECONDS · NO CARD · NO SALES CALL