MCP & Models
Give your AI
the full picture.
Keep Claude, ChatGPT or your own agent. Add Gosset’s connected pharma data and predictive models.
GossetData + modelsThe difference in practice
Same question.
More to work with.
A broader evidence base gives your assistant more of the landscape to reason about.
| Question | Claude | Claude + Gosset |
|---|---|---|
| Which China-origin ADCs now in Phase 2 or 3 have licensing agreements with US partners since 2023? Data pulled · 8 Sept 2026 | 2licensing programs | 4licensing programsAcross 4 distinct ADC assets |
| What’s the competitive landscape of TL1A? Data pulled · 1 Sept 2026 | 9assets | 52assetsPipeline, trials & ownership |
Broader coverage
Find the connections across assets, trials, companies and deals.
Faster answers
Pull relevant records directly, with less searching and extraction.
Less token overhead
Spend fewer tokens gathering evidence and more on the analysis.
Works where you work
One dataset.
Your choice of workflow.
Your AI assistant
Connect Claude or ChatGPT to Gosset and ask questions in the conversation you already use.
Connection guidesCLI & APIYour own workflow
Query pharma records from a terminal, notebook or application. Get structured data you can use.
Start buildingAgentsYour research team
Give your agents and subagents Gosset tools for focused searches, landscapes and diligence.
Teach your agentThe evidence layer
Pharma data.
Already connected.
Gosset structures evidence from registries, publications, conference abstracts, filings and company updates into records your assistant can query.
Explore the dataDrugs & targets
Mechanisms, modalities and development stages.
Trials & results
Design, status, safety and efficacy evidence.
Companies & deals
Developers, counterparties and licensing relationships.
Continuous updates
New evidence linked to the records it changes.
Go beyond retrieval
Evidence in.
Foresight out.
Give your assistant models for the questions a list of records cannot answer.
- Probability of success
- How likely is this trial to succeed?
- Trial timelines
- When could the next milestone arrive?
- Asset valuation
- What does the evidence imply for value?
Forecasts evaluated against observed outcomes, using point-in-time data.
How we measure accuracy
Models are evaluated out of sample against known outcomes. In our trial completion-date evaluation, predictions landed a median of 47 days closer to the actual date than sponsor estimates. Probability-of-success estimates are assessed with calibration curves.