How a brief is made.

What we read.#

Every 24 hours we ingest newly indexed papers from PubMed and registered trials from ClinicalTrials.gov. Before anything reaches your brief it clears an ingest filter: animal-only studies and known predatory journals are dropped, non-research items are excluded, duplicates are merged on DOI or PMID, and retracted papers are removed. We track 36 core journals across the ten ophthalmology subspecialties, though relevance is decided by the classifier, not a journal allow-list. Preprints are included, clearly labeled as not peer-reviewed, and ranked below peer-reviewed work.

How a paper gets through.#

From there it is one fixed nightly pipeline: each paper is read overnight, filtered, classified by subspecialty, evidence-graded, ranked for you, and its summary fact-checked, so your brief lands by 06:00 CET. Every stage runs on a schedule and leaves its own audit trail: reproducible, inspectable, versioned.

How we judge a paper.#

Two deterministic scores travel with every paper, both computed from the structured study data we extract. No language model decides whether a study is strong or whether it is biased.

An evidence level on the CEBM L1 to L5 hierarchy: an RCT outranks a cohort outranks an editorial.

And a deterministic bias inventory across six clinical axes. Each flag fires only when the extracted study data, or a targeted scan of the abstract, says so.

How we summarize it.#

Every summary is the same five-field shape. A headline. A why-now line. A take-away. A one-sentence limitation. A one-sentence prior-evidence framing. The summaries are automated and physician-designed. That is what you see, every weekday.

Nine deterministic checks run on every card before delivery, every number traced back to the source paper. If a check fails, the card is rewritten. If it still cannot be made complete and correct, it is held back, and the next fully-checked paper takes its place. A card you read is never half-built: every one carries a complete scan, or it does not ship.

We don't publish simulated numbers. From launch, validator agreement and the correction rate per thousand summaries are tracked, and published once enough live operation has accrued to be meaningful.

Why you see what you see.#

The morning ten are ranked for you, and for you alone: built from what you read and how you rate it, weighted by the strength of the evidence, never by who paid. No advertiser boost. No paid placement. Ever.

The ranking learns from your account alone, never across readers, and your profile is never a marketing segment. New accounts lean on declared subspecialty and paper quality until there is enough history to personalize. The weights are versioned and audit-logged; every change bumps the version.

When something is wrong.#

We will be wrong sometimes. Our job is to make it rare, catchable, and fixable. Every article carries a Report this summary link. Reports are reviewed as soon as we can, typically within a few working days.

How a correction lands

1. 01

   Editorial sampling.

   Every card clears nine automated checks before it ships. From launch, a rotating sample of shipped cards is additionally reviewed by a clinician against the source, with each outcome (pass, rewrite, correct) logged.

2. 02

   Reader reports.

   Reports go to the same queue. Response window is one working day.

3. 03

   Confirmed corrections.

   The card is regenerated, and the original text is preserved in our audit log, never deleted, so every change stays traceable. Nothing is rewritten in silence.

4. 04

   Quarterly review.

   Subspecialty taxonomy, the curated journal list, ranking weights, validator thresholds, and editorial style guide are reviewed every quarter. Material changes bump the methodology version.