Evaluate Axelium
Resources
Proof and answers for teams evaluating Axelium — a live review you can open now, case studies with the numbers shown, and methodology guides that take a position.
Start here
Live demo · No accountA worked review, open and read-only.The search, the screening decisions, the extracted data and the pooled result — every number traceable to its source.Open the demo
- Product tours · 6 seriesWatch the agents do the work.33 chapters, about 45 minutes end to end: a whole review from a plain-language question to a graded result, then the assistant that answers questions about the evidence.Watch the tours →
- Case studyCan AI reproduce a published meta-analysis?We rebuilt a published review end to end — same searches, same criteria — and compared every number against the original.Read the reproduction →
- ValidationWhat we validate — and where the limits are.The standing account of automation boundaries, reviewer responsibilities, and known limitations.Validation & limitations →
Guides
Methodology questions, answered directly
Every guide opens with the short answer, then explains what it depends on and where it stops being true — written to be defended at peer review.
Effect measures4 guides
Models & heterogeneity2 guides
Network meta-analysis2 guides
Certainty & bias2 guides
Question design & process2 guides
Field notes
Featured · MethodologyWhat an automated review must not decide.The useful question is not how much of a review a system can run unattended. It is which decisions it may make on its own — and which must be locked before the run begins. On task autonomy versus control autonomy, why a system that re-runs a stage until the result improves is an automated p-hacker, and what has to stay human.Read the resourceMethodology · 05Locked before the run begins.Notes from the field
Research notes and case studies from building — and stress-testing — AI evidence synthesis.
Earlier in the archive
- 6 July 2026When the indirect comparison is the weak linkEvidence note
- 22 June 2026Should AI read every paper, or filter first?Research note
- 14 May 2026The Three-Layer Pattern: LLMs Propose, Algorithms Decide, Humans ValidateMethodology
- 13 April 2026Can AI reproduce a published meta-analysis? We tested it.Case study
Evaluating Axelium?
Bring the protocol you need to defend.
We’ll walk through how Axelium supports your review question — methods first, procurement paperwork when your institution needs it.
Using Axelium already?
Setup, collaboration, billing and the methodology notes live in the documentation.