Skip to content

Clinical evidence, brought into order

The literature is scattered.
The review is not.

CERULEON gathers the clinical literature from PubMed, imports other databases, screens publications against PRISMA 2020, grades it by Oxford CEBM and GRADE, and structures it into a ready-to-use Literature Search Report — so your evaluation starts from a map, not a mountain of PDFs.

See the workflow

Free tier includes the full manual workflow. No card required.

Gathered — the whole literature, every database.

Screened — against PRISMA 2020, each study kept or excluded, with a reason.

Graded — weighed, not just counted: Oxford level, GRADE confidence.

Structured — one outline out, sourced and graded, ready to write.

AI-drafted sections — sign-off rests with the qualified evaluator.

02Why this exists

A clinical evaluation shouldn't cost you a quarter of the year.

Three structural problems make most reviews late, shallow, or both. CERULEON addresses each one directly.

01

1.5Mnew PubMed citations a year

Paper volume outpaces reading speed

A typical clinical evaluation pulls 800 to 2,000 abstracts before exclusion. Excel and PDF folders don't scale beyond a few hundred references before structure breaks down.

02

25–75 hto screen a 1,500-record pool

Manual screening costs weeks

Title-and-abstract screening with two reviewers takes 30 to 90 seconds per record — before inter-rater calibration. Most small teams cut corners and ship under-evidenced reviews.

03

4audit artefacts Word won't enforce

The output isn't audit-ready

Notified bodies, ethics committees, and journal editors want PRISMA flow diagrams, a documented evidence-grading method (such as Oxford CEBM levels), exclusion logs with reasons, and per-claim source citations. The structure has to come from the tool.

03What CERULEON does

From a scattered field to a structured report.

03.1Search

Search PubMed the way reviewers actually do.

The query editor maps to PubMed's E‑utilities and respects MeSH explosion, field tags, and Boolean precedence. Save queries per project, re-run them after the literature has moved, and dedupe against your existing pool automatically. The query history is part of the audit trail.

Bringing references from elsewhere? Drop in a file. CERULEON parses RIS, NBIB, BibTeX, and Cochrane / Europe PMC exports — and de-duplicates against everything already in the pool by DOI, PMID, or normalized title.

  • MeSH explosion, field tags, and AND / OR / NOT precedence
  • Multi-source import: RIS, NBIB, BibTeX, manual paste
  • Auto-deduplication across queries, imports, and prior pools

03.2Screen

Screen 200 abstracts per minute. Read the ones that matter.

Auto-screening lets you screen articles with AI against your inclusion and/or exclusion criteria. Every decision returns the model's reason, quoted from the abstract. You can verify the logic in seconds instead of re-reading the paper.

You stay in the driver's seat. Bookmark anything ambiguous. Manually override any AI call — every override is timestamped in the audit trail. Lock the pool when the screen is done so subsequent edits are tracked.

  • Full-text PDF upload, linked to the corresponding publication for easier access
  • Quoted source justification on every include / exclude
  • Manual override on every decision, fully logged

03.3Classify

Oxford and GRADE levels, computed from the design.

CERULEON identifies study design from title and abstract and assigns an Oxford Centre for Evidence-Based Medicine level (Ia – IV). For systematic reviews of randomized trials it also applies a GRADE judgement: High, Moderate, Low, or Very low confidence.

The dashboard pyramid shows your evidence base at a glance. Drill into any tier to see the underlying studies. Re-run classification after each round of new inclusions — it's idempotent and version-stamped, so re-classifying never changes your audit history.

  • Oxford CEBM Ia / Ib / IIa / IIb / III / IV
  • GRADE confidence rating: High / Moderate / Low / Very low
  • Pyramid view with drill-down to underlying studies

03.4Tabulate

Every included study, distilled into one comparable table.

CERULEON generates a structured evidence table from your included pool — one row per study. Objective, population, key results, and safety outcomes line up in columns, so the whole body of evidence reads at a glance instead of staying buried across dozens of abstracts.

The table is built from the data already in your project and stays in sync as you include or exclude records. It lands in the PDF report and exports standalone as XLSX for offline analysis or document integration.

  • One row per study: objective, key results, and safety side by side
  • Includes citations of the articles' extractions
  • Full AI analysis through uploaded PDF publications
  • Add your own AI extraction columns — define a field once, it's extracted across every study

03.5Synthesize

A synthesis you can defend, with sections you can rewrite.

CERULEON generates a structured synthesis split by clinical theme — efficacy, safety, study quality, evidence gaps — with inline citations to your included pool. Each section is editable inline. Edit a paragraph, regenerate just that block, or paste your own conclusions.

Every citation is verified against your included pool. A built-in check catches any reference the AI invents or that points outside your bibliography and removes it — so no claim ever rests on a source that isn't there. And when the standard chapters don't fit your review, add your own with a custom prompt; it follows the same citation rules as every other section.

  • Sections: efficacy, safety, study quality, evidence gaps
  • Per-section regenerate without losing your edits
  • Every citation verified against your included pool — invented or out-of-range references caught and removed
  • Custom review mode — add your own chapters, each with its own prompt and order

03.6Export

One click. Three audit-ready deliverables.

One click turns your whole review into a finished, typeset PDF — laid out to Swiss medical reporting conventions, with everything a reviewer expects already in place: PRISMA flow diagram, evidence pyramid, included-studies table with PICOTS scoring, GSPR mapping (Annex I, Ch. I, items 1–9 in Device mode), full bibliography, a reasoned exclusion log, and the complete AI audit trail. It's built to annex straight into your Clinical Evaluation Report — you keep authorship, and regulatory sign-off stays with your qualified evaluator. Prefer to keep editing? Export the very same report as LaTeX.

Only need one piece? Export any module on its own, in the format that fits the job — XLSX for offline analysis, TXT for quick stakeholder edits, RIS for your reference manager, or PDF for a clean standalone. Every file carries the same project version stamp, so your deliverables always line up, whatever the format.

  • Literature search report exported to PDF or LaTeX
  • Individual modules exported to the right format: XLSX, TXT, RIS, PDF
  • Version stamp on every deliverable for cross-reference

04From query to PDF in four passes

A workflow that mirrors how reviews are actually done.

Search, screen, synthesize, export. Each step has a clear handoff and a clear deliverable. No back-and-forth between five different tools.

Search

Configure your PubMed query or import an existing pool from RIS / NBIB / BibTeX. CERULEON dedupes and stores everything in your project.

Screen

Run AI screening with your inclusion criteria. Review the model's reasoning. Override anything you don't trust. Lock the pool when done.

Synthesize

Generate the structured synthesis. Edit any section inline. Regenerate sections individually. Lock the version when ready for review.

Export

Download the PRISMA-2020-structured literature review as PDF, plus XLSX, TXT, RIS, or LaTeX. Every deliverable carries the same version stamp.

05Pricing

Pick the tier that matches your throughput.

Start free with the full manual workflow. Upgrade for AI screening, deeper synthesis, and bigger PDF storage. Cancel any month, no contract.

Free
Try the workflow without AI.
0 EUR forever
  • PubMed search + import (RIS/NBIB/BibTeX)
  • Manual screening + PRISMA flow
  • PDF/XLSX/LaTeX exports
  • 5 PDFs storage
  • No AI features
Pro
AI-assisted reviews for solo consultants.
19 EUR / month
  • Everything in Free
  • Basic AI features (Table extraction / General Synthesis)
  • 100 PDFs storage
  • Medium AI limit
Most popular
Pro+
Power tier with Heavy Mode access.
49 EUR / month
  • Everything in Pro
  • Deeper AI reasoning
  • Auto-screening with GRADE evaluation
  • 500 PDFs storage
  • Extended AI limit
Pro Max
For heavy users — large reviews, big projects.
199 EUR / month
  • Everything in Pro+
  • 2000 PDFs storage
  • Maximal AI limit
  • Priority support

Prices shown in EUR. CHF and USD billing detected automatically at checkout. Yearly billing gives you 2 months free (pay 10× the monthly price). VAT added where applicable.

06Common questions

Eight answers before you ask.

Where do the source records come from?

PubMed (via the official E-utilities API) is the primary source. You can also import records from Cochrane, Europe PMC, Embase, or your library's RIS / NBIB / BibTeX exports. Every imported file is de-duplicated against your existing pool by DOI, PMID, or normalized title.

Is the synthesis PRISMA-compliant?

Yes — the report follows the PRISMA 2020 reporting structure. CERULEON auto-generates the PRISMA 2020 flow diagram from your search counts, deduplication counts, screening decisions, and inclusion outcomes. The exclusion log carries the AI's reason or your manual reason for every excluded record. The included studies table maps to the PICOTS framework. All of this lands in the exported PDF.

How does CERULEON handle data privacy?

CERULEON only processes published abstracts and PDFs you upload. No patient data is collected or processed. Uploaded PDFs are stored encrypted in Supabase Storage and never sent to any AI model unless you trigger an action that requires it (e.g. full-text screening on Pro+).

The application is hosted in the EU (Frankfurt region).

Which export formats are supported?

PDF (full literature search report or individual modules), XLSX (full extraction tables for document integration or offline analysis), TXT (for stakeholder review and editing), RIS (for bibliography management), and LaTeX (for open source editing). All exports carry the project version stamp.

Can I cancel anytime?

Yes. Monthly plans cancel at the end of the current billing period, with no contract and no cancellation fees. Yearly plans remain active through the end of the paid term. Your data stays in your account either way; you keep read-only access for as long as the account exists.

What's actually in the free tier?

The complete manual workflow: PubMed search, RIS / NBIB / BibTeX import, manual screening with PRISMA flow, full extraction table, evidence pyramid (computed from your manual study-design tagging), and PDF / XLSX / LaTeX export. No AI screening, no AI synthesis, and storage capped at 5 PDF uploads. Good enough to run a small review end-to-end without paying.

Do you offer institutional or team plans?

Multi-seat plans, SSO (SAML / OIDC), and on-premises Supabase deployment are available for hospitals, CROs, and notified bodies. Contact us at contact@ceruleon.app with your team size and we'll come back with a quote within 48 hours.

The literature review, minus the all-nighter.

Free tier covers the full manual workflow. Spin up your first project in under a minute. No card, no setup, no consultant call.