Skip to content

Research changed. Optimize for what reads it now.

Trace how evidence travels.

Your paper is now read twice — once by reviewers, then forever by machines summarising it. Trace rehearses both: a simulated editor and reviewer panel before you submit, and measurement of how answer engines find, cite, and represent you after you publish.

Analyze a paperSee how it works

Built for researchers, publishers, institutions, and scientific teams.

The gap

Cited is not the same as understood.

A paper can be cited constantly and still have its effect size restated as something it never said. Trace measures those separately, because they fail separately.

27%

of the questions this paper should be able to answer end with it used and faithfully represented.

of 1,240 benchmark questions · TVI_v0.1

  • Relevant questions100%
  • Retrieved62%
  • Cited43%
  • Evidence used31%
  • Interpreted faithfully27%

Each stage measured independently · demo workspace

Two doors, one engine

Survive review. Then survive the summary.

Whatever makes a paper hold up under a reviewer is most of what makes it hold up being summarised by a machine. Trace runs both from the same extraction.

Before you submit

Students, first authors, labs

Get it accepted.

A simulated editor-in-chief returns a decision and the criteria behind it. Three reviewers with different expertise write reports anchored to the sentence each is objecting to. The reporting standard your design requires is checked item by item, and the manuscript is measured against the author instructions of the journals you are considering.

  • Editor decision, criteria scores, and desk-reject risk
  • Reviewer reports pinned to the passage that provoked them
  • CONSORT and reporting-standard coverage, item by item
  • Journal fit, and what each venue would want changed first

82%

CONSORT coverage on the seeded trial

Run a review

After you publish

Institutions, publishers, life sciences

Get it found, and read correctly.

Discovery runs through generative answer engines now, and they are not search. Trace measures retrieval, citation, evidence use, and interpretation fidelity separately, finds what appears instead of you, and turns the gaps into metadata, availability, and terminology work — the optimisation layer for AI-mediated discovery.

  • Retrieval, citation, and evidence use measured independently
  • Interpretation fidelity across twelve dimensions
  • The substitutes that appear in your place, and why
  • Longitudinal series with confidence bands, re-measured on a schedule

62 → 27%

retrieved, to faithfully represented

See the measurement

Editorial review is modelled output · journal templating is on the roadmap

What Trace optimizes

Three surfaces, not one dashboard.

Measurement alone is a scoreboard. These are the three things the work changes.

01The paper

Written to get past review, and past a summary.

A simulated editor-in-chief and three reviewers return the decision before a journal does. Every suggested edit is checked against the extracted claims and discarded if it would change what the paper asserts.

0

suggestions that alter claim strength

Run a review
02The reach

Found by the systems people actually ask.

Discovery runs through answer engines now, and they are not search. Trace measures retrieval, citation, and evidence use separately, then finds what is keeping you out.

62 → 27%

retrieved, to faithfully represented

How measurement works
03The reception

Stress-tested before a clinician reads it wrong.

Panels of specialists, payers, and patients read the paper and deliberate, surfacing the misreadings it invites while there is still time to rewrite.

23%

read the relative effect as absolute

How panels work

Demo workspace figures · no ranking promised

One paper, all the way through.

01 / Discover

See what AI sees.

Every paper becomes a benchmark of the questions it should be able to answer. Trace measures where it appears, what replaces it, and whether the evidence is actually used.

34%of relevant questions use its evidence
Explore Discover
  1. Discover
  2. Simulate
  3. Improve
  4. Launch
  5. Monitor

Discover: See what AI sees.

Every paper becomes a benchmark of the questions it should be able to answer. Trace measures where it appears, what replaces it, and whether the evidence is actually used.

34% of relevant questions use its evidence

Simulate: Read the paper through another mind.

Synthetic stakeholder panels surface likely misreadings and objections before expensive human research. Every output is labelled simulated.

27% read the relative effect as the absolute one

Improve: Easier to find. Identical in meaning.

Terminology gaps, ambiguous abstracts, and missing structure. Any suggestion that would strengthen a claim is discarded rather than shown.

0 suggestions that change claim strength

Launch: Publication is a beginning.

Metadata, availability, institutional presence, and attributable author communications — each with a disclosure and an approval step.

23 availability and metadata checks

Monitor: Then measure it again.

Visibility changes. When an open-access deposit breaks, the availability score falls and the regression is flagged — recorded as having occurred, never as having been caused.

5 measurements over 8 months

Methodology

A score you can inspect.

Every index is built from observable runs, versioned query sets, and source-level evidence. Ordering is public; the formula ships with the product.

Trace Visibility Index

TVI · TVI_v0.1

Retrieval
relative emphasis Highest
Citation
relative emphasis High
Evidence use
relative emphasis Supporting
Fidelity
relative emphasis Supporting
Breadth
relative emphasis Supporting
Stability
relative emphasis Supporting

Availability carries a supporting weight. Components are always reported alongside the composite, because one number cannot tell you whether a paper was never retrieved or retrieved and then misread.

The Trace Visibility Index is a product heuristic under active validation, not a validated scientific metric.

Who it is for

One paper, or four hundred thousand.

The loop does not change with scale. Only who approves what, and which failures are worth fixing centrally.

  1. 01

    Researchers and labs

    One baseline, and a list of defects you can actually fix — metadata, terminology, an abstract that truncates badly.
    For researchers →
  2. 02

    Universities and research offices

    Which work is visible, which is under-discovered, and where a deposit or identifier gap is doing the damage — by department, not paper by paper.
    For institutions →
  3. 03

    Publishers and journals

    How the catalogue surfaces by journal, imprint, and subject — and the availability failures holding back whole segments at once.
    For publishers →
  4. 04

    Life sciences and medical affairs

    How published evidence is being represented, stress-tested before launch, with scientific exchange and promotion kept separate and both on a review trail.
    For life sciences →

Most people will meet your research through something else’s summary.

Trace rehearses both readings — the reviewer’s and the machine’s — before either one is final.

Open the sandboxTalk to us

The sandbox walks the whole loop on a seeded fictional trial — measurement, simulation, revision, launch, and monitoring — with every number replayed from a recorded run. No account, nothing uploaded.

Trace never promises a ranking improvement, and does not model how any engine ranks sources.