For pharma R&D — target assessment and portfolio

QIW, by Entreater, helps R&D and portfolio teams rank drug targets on traceable evidence

— and shows when the evidence cannot yet name a clear winner.

See a sample decision Talk about your use case

Your portfolio needs established targets, emerging ones and some that are entirely new — how sure are you about the order within each?

See examples

Finished runs to explore: ranking, judgements and evidence without signing in.

Four NSCLC drug targets, ranked from evidence

A small, real run: four non-small cell lung cancer drug targets (ALK, EGFR, KRAS, MET) compared pairwise on 20 criteria by one LLM judge, using only the evidence dossiers collected beforehand. 240 judgements, about a fifth of them abstentions where the dossiers held no usable evidence. The ranking is a Bayesian Bradley-Terry-Davidson posterior with intervals; every judgement, rationale and cited source is open below. Not a clinical recommendation.

4 candidates

240 judgements

22.1% abstention

Published on Sep 15, 2026

Open example

See all examples

Why this is different

In our public example of four NSCLC targets, ALK and MET form a front group, separated from KRAS and EGFR. Within each group, QIW says: not decided.
The judge has no web search, it has to cite the evidence it relied on, and every pair can run in both orders, so position effects become visible.
Thin evidence shows up as a wide interval, and judges abstain and say why. A candidate cannot reach the top by being unknown.

The films

How sure are you?

QIW, by Entreater: for R&D and portfolio teams who need to know how sure a target ranking is.

Ranking under uncertainty

QIW, by Entreater — for data scientists and evaluators: when is first place in a ranking really separated?

The controlled judge

QIW, by Entreater — for evaluators of LLM judges: which controls turn a model's verdict into usable evidence.

See a sample decision Talk about your use case

Full lecture