Now analyzing 2.3M+ papersBeta

Citations Measure Attention.We Measure Evidence.

The due diligence platform for science.

2.3M

Papers Indexed

47M

Claims Extracted

156K

Contradictions Found

The Unexamined Hypothesis

Science has a
reproducibility problem

Socrates taught that wisdom begins with admitting what we don't know. Yet science builds on claims that have never been questioned— citations measure attention, not truth.

0%

of preclinical research fails to replicate

Nature, 2016

0B

wasted annually on irreproducible research

Freedman et al., PLOS Biology

0M+

papers published every year—impossible to verify manually

NSF Science & Engineering Indicators

"The unexamined claim is not worth building on."

We built search engines for finding papers. Now we need systems for questioning them.

The Socratic Method

Truth through
relentless questioning

Socrates exposed truth by asking the right questions. We do the same for science— interrogating every claim against the evidence.

1

Question

Interrogate every claim

Like Socrates in the agora, we dissect papers into testable assertions. Not abstracts. Not conclusions. The actual claims that demand examination.

2

Dialectic

Trace the dialogue

Every claim exists in conversation with others—replications, contradictions, refinements. We map the dialogue so you see truth emerge.

3

Elenchus

Expose contradictions

The Socratic method reveals what doesn't hold up. We score each claim by evidence strength—replications, methodology, consistency—not popularity.

The result? Examined claims with evidence-based confidence scores.

Know what survives scrutiny before you build on it.

See it in action

From paper to claims to evidence — the complete pipeline.

app.socratic.science/papers/10.1038-s41591-023-02847-x
Research Article

TGF-β Pathway Inhibition by Compound X Reduces Hepatic Fibrosis in Murine Models

Zhang, L., Chen, M., Williams, K., et al.Nature Medicine2023

Paper Score

74

5 claims

Extracted Claims

Click a claim to explore

Built for those who
demand examined truth

When decisions cost millions (or lives), you need answers that survive questioning.

Early Discovery Teams

Pharma R&D

Validate drug targets before committing years and billions. Know which preclinical findings actually hold up.

⚠️ Bad target selection costs $1B+ and a decade of work.

  • Evaluate target validity before project commitment
  • Surface replication failures across compounds
  • Track mechanism claims across the literature
  • Reduce false starts in early discovery
Get Started
"We spent three years on a target that turned out to be based on irreproducible findings. Never again."

VP Discovery, Top 10 Pharma

We don't accept.
We interrogate.

Citation databases count who's listening. Expert networks collect opinions.
We ask the questions they don't—and show you what survives.

Socratic
Citation DBs
Expert Networks
AI Search
Extracts specific claims from papers
Methodology quality flags
Tracks replication attempts
Surfaces contradictions automatically
Evidence-based credibility scoring
Retraction tracking
Real-time updates as new papers emerge
Scales to millions of papers
API access for pipelines

The Socratic difference

Every claim examined

Like Socrates questioning each assumption, we isolate and evaluate claims independently. No assertion escapes scrutiny.

Evidence over authority

Socrates challenged the experts of his day. We challenge citations with replication data—truth isn't a popularity contest.

The dialogue continues

Knowledge evolves through ongoing examination. Our scores update as new evidence enters the conversation.

The Science Graph

The dialogue of
evidence

Every claim in conversation with every other. A living dialectic of what science has examined—and what survives questioning.

0

Papers Examined

0

Claims Questioned

0

Dialectic Links

0

Contradictions Exposed

Real-time updates

New papers are processed daily. Scores update as evidence emerges. The graph never sleeps.

Cross-domain links

A finding in oncology might support or contradict one in immunology. We connect the dots.

API-first design

Build your own tools on top of the Science Graph. Structured data, not PDFs.

Built by people who
lived the problem

One of us spent years making clinical decisions on unreliable evidence. The other built systems to find signal in noise at massive scale.

Janny Ke, MD

Janny Ke, MD

Co-founder

Anesthesiologist turned epidemiologist. Harvard MS in Epidemiology. Spent a decade watching clinicians make decisions on shaky evidence. Founded AnesEd.ca to make medical education more rigorous.

MD, Harvard Epidemiology MS
Providence Health Care
AnesEd.ca Founder
Ning Ma

Ning Ma

Co-founder

Former Head of Data Science at Checkr and Rocket Fuel. Angel investor in Anthropic, SpaceX, and 20+ frontier companies. Holds an AI patent for dynamic content optimization—predicting the next page, not the next token.

Ex-Checkr, Ex-Rocket Fuel
AI Patent Holder
UC Berkeley

Advisors

Coming soon

Clinical Advisor

Top Academic Medical Center

Coming soon

Science Advisor

Major Pharma R&D

Coming soon

Technical Advisor

Leading AI Company

Founders backed by the same investors as

AnthropicSpaceXCerebrasShield AIGeneral Fusion

Join the pursuit of truth

The examined life requires examined evidence. Get early access and join the dialogue.

No spam. Just updates on our progress and your invite.

Trusted by teams from

PfizerGenentechFlagship PioneeringARCH VentureRecursion