Quick
free to startDrop a file. Get a map.
Evidence maps, in motion
Where research is dense, where it is thin
Examine Evidence is a web app for evidence synthesis: you search or import studies, screen and classify them with rules, AI reviewers and people, and build evidence maps that show where research is dense and where it is thin. Every decision, by a person or by AI, is recorded.
Your key is encrypted at rest and never shown again.
01 Import
Any file. Columns mapped for you.
The table appears at once.
02 Screening
Excluded studies sink, never vanish.
03 Classify
Classes by rules, AI or people.
04 Map
Hover a cell. Click to pin it.
| Exercise | Diet | Drug | Education | Digital | Surgery | Combined | |
|---|---|---|---|---|---|---|---|
| Mortality | 12 | 9 | 42 | 3 | 0 (gap) | 18 | 7 |
| Quality of life | 38 | 24 | 15 | 20 | 11 | 0 (gap) | 29 |
| Pain | 46 | 5 | 33 | 8 | 0 (gap) | 27 | 14 |
| Cost | 0 (gap) | 7 | 11 | 0 (gap) | 9 | 6 | 0 (gap) |
| Adherence | 21 | 17 | 26 | 34 | 40 | 0 (gap) | 12 |
06 Validate
Example agreement matrix with synthetic values: of 100 sampled studies, the AI and you both include 42 and both exclude 45; you include and the AI excludes 8; you exclude and the AI includes 5. Agreement 87 percent, kappa 0.74. These numbers are an example, not a result.
Every AI decision, visible
Failures are flagged, never silent
Every decision on record
Your data stays yours
Drop a file. Get a map.
Search, screen and classify in one place.
Contact us for pricing Add your model key and get started nowYour key is encrypted at rest and never shown again.
An agent works. Every step saved.
Prepare evidence maps for free in minutes
Make a mapNo per-seat AI subscription. Pay for what the AI actually uses, or bring your own key and pay the provider directly.
Example map, synthetic data. No real studies.
Open the example mapAbout
Examine Evidence is a web app for evidence synthesis: you search or import studies, screen and classify them with rules, AI reviewers and people, and build evidence maps that show where research is dense and where it is thin. Every decision, by a person or by AI, is recorded.
An evidence map, also called an evidence gap map, is a table that counts the studies on a topic by two classifications, for example outcomes in rows and interventions in columns. Dense cells show where much research exists. Empty cells show possible gaps. A map shows where the evidence is; it does not say what the studies found.
People and teams who need to make sense of a body of literature, for example to build an evidence gap map or a scoping review. Quick is for one person who wants a map fast. Studio is a full workspace for teams that search, screen and classify together.
Quick is free to start. Sign in with your email address, make a map and share it. A free account includes a small monthly AI budget. Studio is priced on request, and Agentic is in beta.
You define the question and the criteria. Rules and AI reviewers read each study and propose include or exclude, with a reason that is saved. To validate, take a sample, mark it yourself, see where the AI disagrees with you, adjust the instructions, then run it on the full set.
In Quick and Studio, no. AI proposes and people decide, and you can override any AI answer. Excluded studies are kept, not deleted. If an AI call fails, the study is flagged as needing review instead of being guessed. Agentic mode, in beta, lets an agent work for you, and every step it takes is saved so you can check it.
RIS, CSV, Excel, BibTeX and JSON files, and Zotero exports. Columns are mapped for you, and the table appears at once. You can also search PubMed, OpenAlex and Semantic Scholar from inside the app.
On a server we run with Hetzner, a hosting company, and backed up to separate storage. The server is in Nuremberg, Germany, behind Cloudflare; see the privacy notice.
Yes. Add your own provider key and Studio opens straight away; you pay the provider directly, with no markup, and the small monthly AI budget of the built-in agents stays. Your key is encrypted at rest and never shown again.