Aibstracts
Start a review
Methodology & rigor

Fast is only worth it if the result is defensible

Aibstracts is built so a methodologist can interrogate every output and a peer reviewer cannot fault the trail. The speed comes from automation. The trust comes from transparency, established frameworks, and you in the loop.

  • Four layers, and only one of them is a model.
  • Published decision rules set the ratings, run as deterministic code.
  • You stay the reviewer of record and own the sign-off.
How the appraisal is actually produced
  1. Layer 1

    Inputs

    • Source PDFs
    • Search results
    • Screening data
  2. Layer 2

    AI proposes

    • Reads the source PDF
    • Answers each signalling question
    • Anchors every answer to a quote
  3. Layer 3

    Published rules decide

    • RoB 2 / ROBINS-I algorithm
    • Deterministic code, not a model
    • Domain and overall judgments
  4. Layer 4

    Reviewer signs off

    • Override any answer
    • Judgments recompute live
    • Calibration against gold standard

Result An auditable, source-anchored, reproducible appraisal.

In detail

Eight pillars of a result that holds up

  • Cochrane RoB 2

    The published RoB 2 algorithm sets every rating, including cluster and crossover trials.

  • ROBINS-I

    Non-randomized studies appraised end to end, routed by study design automatically.

  • Source anchoring

    Every answer carries the source quote, one click away in the PDF.

  • Reviewer verification

    Override any answer and the judgments recompute live. You sign off.

  • Calibration

    Agreement against a reviewer-validated gold standard, reported as a number you can show.

  • PRISMA provenance

    Your PRISMA flow built from the real trail, never reconstructed at the end.

  • Evidence Profiles

    Studies outside RoB 2 and ROBINS-I get a structured profile, never dropped.

  • Reconciliation

    Full-text conflicts with the abstract stage are surfaced, not buried.

What we don’t claim

Honesty is part of the method

  • It does not replace the reviewer. The methodological accountability and the sign-off stay with you.
  • It is honest about scope. Protocol, search, screening, and appraisal are in production; synthesis support is still evolving.
  • It shows its work. Every model-read judgment carries the quote and its location, never a bare confidence score.
  • It is transparent. Models propose the evidence; the published RoB 2 and ROBINS-I rules produce the ratings.
For teams whose evidence has to hold up

Put the methodology to the test

Run a review you have already done by hand and compare. Calibration will tell you exactly how close it gets.

Book a demo

A free plan is available, no card required