Discover what’s hidden in existing science.
GXL builds the knowledge layer AI agents need to reliably search and connect scientific evidence at scale. Our AI-native indexing makes it possible for agents to quickly and reliably identify therapeutic hypotheses worth testing.
More of what science already knows, for the questions that come next.
The next therapeutic lead may be hiding in plain sight. Your AI agents need a structured way to find it.
The scientific record is distributed across generations of papers, databases, clinical trials, regulatory records, figures, supplements, and accumulated research. WIth AI, agents can search and synthesize this data at a scale no one has seen before. AI helps us draw connections we couldn’t see, to develop new therapies for patients.
GXL gives AI agents one structured environment to connect scientific findings across sources and preserve the path back to the evidence.
Faster, more reliable investigation can make more therapeutic ideas feasible to pursue, and reduce risk across the development lifecycle.
-
Broader evidence
Search across papers, patents, trials, regulatory records, and biological databases in a single investigation.
-
Faster investigation
Explore multiple evidence threads in parallel without losing the path back to the source.
-
Lower compute cost
Persistent indexed context and agent-native queries cut token spend per task and across repeated runs
-
Source level traceability
Every synthesized result can preserve the route back to the underlying evidence.
GXL's Paperclip is is trusted by
- 12 of the top 20 pharma companies
- 20 leading biotechs
- 100+ universities and research institutes
In one FDA analysis, an AI agent with GXL’s Paperclip flagged 18 of 21 drugs—2.1× faster at 47% lower cost.
Read moreEvaluated on scientific work
Paperclip performance on bioRxiv questions
Average across 140 real research questions: deep paper Q&A (50), experiment novelty checks (50) and multi-paper synthesis (40).
Read more- 91%
- 57%
- 31%
- 14.9m
- 4.5m
- 1.9m
- $0.37
- $1.32
- $1.00
- Paperclip Chat
- Claude Code + bioRxiv MCP
- FutureHouse Edison
Paperclip performance vs other agents across 20 protein questions
20 questions spanning target characterization, structural profiling, and druggability assessment
Read more- 100%
- 80%
- 85%
- 55%
- 40%
- 97.1%
- 78.3%
- 81.9%
- 53.8%
- 39.5%
- 1.7m
- 6.5m
- 12m
- 9.1m
- 13m
- $0.22
- N/A
- N/A
- $1.24
- $1.93
- Paperclip
- Phylo
- Claude Science
- MCP
- Web Search
Paperclip performance on FDA analysis
Replicating one published FDA analysis: 18 of 21 drugs flagged, 2.1× faster and 47% lower cost with Paperclip.
Read more- 86%
- 71%
- $0.45
- $0.85
- 65 s
- 136 s
- With Paperclip
- Without Paperclip
Generate and analyze novel datasets to form new hypotheses and de-risk development.
-
Therapeutic research
Examine targets, mechanisms, and disease biology for leads worth testing.
-
Risk assessment
Check prior trials, patents, and regulatory findings before the next decision.
-
Translational science
Connect preclinical findings with clinical and biological context.
-
Clinical development
Follow related mechanisms, programs, and patent activity across a therapeutic field.
-
Competitive intelligence
Follow related mechanisms, programs, and patent activity across a therapeutic field.
-
Drug repurposing
Investigate new uses for known compounds against existing evidence.