Great science has a signature.
The world’s clinical evidence in your own AI assistant: every registered trial, the drugs and their interactions, the literature and the regulatory guidance, every answer with its source. And the connection to the pharmaceutical companies able to develop your research, with your name on the work.
Free for researchers and institutions. Every tool is opening to registered researchers in Claude and ChatGPT, through MCP.

The evidence tools
The world’s clinical evidence, in the assistant you already use.
MCP is the open standard by which an AI assistant uses tools. Register once, add the MondoTwin connector to Claude or ChatGPT, and your assistant gains the tools of the twin: it searches every registered trial, reads a trial in full, screens a drug pair, looks up a disease’s targets and opens the ICH and FDA guidance. Your assistant asks; the twin returns the records with their sources; the reasoning happens in your conversation.
What the twin holds
- Every registered trial across six registries, a trial registered twice reduced to one record
- The drugs, with mechanism, targets, metabolism and interactions
- The biomedical literature, linked to the trials it reports
- The rare-disease entries, anchored to genes and targets
- The ICH guidelines and the FDA guidance, searchable by passage and readable in full
- Patents, connected to the molecules and the trials
See it answer
Five questions, answered with the records.
Five example questions asked through Claude with the MondoTwin connector. For each, the tools the assistant calls, what each returns, and the answer with its sources.
Has adding sildenafil to nintedanib been tried in idiopathic pulmonary fibrosis, and is the pair safe to give together?
- screen_ddisildenafil, nintedanib · chronic exposuremonitor · medium confidence · 2 curated records · no shared enzyme · haematologic flag on both
- search_clinical_studiessildenafil nintedanib · idiopathic pulmonary fibrosisINSTAGE, NCT02802345 · phase 3 · completed · 274 patients · negative result
- get_clinical_studyNCT02802345arms, eligibility, endpoints, registry results, 8 publications
- check_fdc_feasibilitysildenafil, nintedanibsingle tablet · moderate · oral · half-lives 4 h and 9.5 h
It has been tried, and it did not help. INSTAGE, Boehringer Ingelheim’s 24-week phase 3, randomised 274 patients with IPF and advanced impairment (DLCO ≤ 35 % predicted) to nintedanib with sildenafil or nintedanib with placebo, in 13 countries. On the primary endpoint, the change in SGRQ total score at week 12, the adjusted mean difference was −0.52 (95 % CI −3.33 to 2.30, p = 0.72). Dyspnoea moved no differently either.
- Registry
- NCT02802345 · EudraCT 2015-002619-14
- Design
- Randomised, double-blind, nintedanib + sildenafil against nintedanib + placebo
- Population
- IPF, 40 years and older, DLCO ≤ 35 % predicted
- Primary endpoint
- SGRQ total score, change at week 12
- Result
- −0.77 against −1.28; difference −0.52, p = 0.72
- Publications
- 8 on PubMed
Given together: monitor. DrugBank records that sildenafil can raise the serum concentration of nintedanib. The two share no enzyme where one is a perpetrator and the other a substrate. Both carry a weak haematologic flag on their labels. Renal effects were assessed for nintedanib only, cardiac rhythm and sedation for sildenafil only.
One product is possible. Both are oral. Half-lives differ 2.4 times (4 h against 9.5 h), so a shared schedule needs a modified-release form of sildenafil. Formulation difficulty: moderate.
SourcesNCT02802345EudraCT 2015-002619-14DrugBank DB00203DrugBank DB09079OnSIDESDILIrank
Example for research use. Medical decisions belong with a physician.
For clinical research
Ask in your own words. The twin answers with the records.
Six things researchers and clinicians do with the twin, and three more. The conversations above show five of them.
- 01
Map a disease
“What has been tried in idiopathic pulmonary fibrosis, by phase and outcome, and where is the evidence thin?”
Every registered trial in the indication, with phase, status, sponsor, enrolment and readout where results exist.
- 02
Follow a molecule
“Which trials tested metformin outside diabetes, and what did they show?”
The trials across indications, one record per trial across registries, with results, sponsors and the registry link for each.
- 03
Design a trial
“Which primary endpoints, comparators and enrolment did phase 3 trials in idiopathic pulmonary fibrosis use?”
Arms, endpoints, eligibility and enrolment reached against enrolment planned, trial by trial: the precedents a protocol cites.
- 04
Check a combination
“Can pirfenidone and nintedanib be given together, has the pair been tried, and could they be one product?”
An interaction screen with its evidence by layer, the trials that tested the pair in any indication, and whether the two actives can share a formulation.
- 05
Ground a mechanism
“Which targets carry genetic evidence in idiopathic pulmonary fibrosis?”
The associated targets from Open Targets, filtered on genetic association, and the pharmacology of each drug you consider: enzyme roles, interaction load, patent and generic status.
- 06
Read the guidance
“Which ICH and FDA guidance governs an interim analysis and an adaptive design?”
The relevant passages of the ICH guidelines and the FDA guidance documents, and the full text of any of them on request.
- 07
Score a dose-response surface
A checkerboard of two drugs comes back with its excess over Bliss, Loewe, HSA and ZIP, and the MuSyC interaction parameter with a bootstrap interval.
- 08
Read your documents with the evidence
Upload a protocol, a grant draft or a manuscript to your workspace and ask across it and the registered trials in one question.
- 09
Keep a working set
Shortlists, triage marks and notes written from the conversation are saved to the platform and appear in your workspace.
Every answer with its source
Records, with their provenance. The judgement stays with you.
One record per trial
A trial registered on ClinicalTrials.gov, EU CTR, CTIS, UMIN, ANZCTR or ISRCTN is one record, carrying every identifier it was given. FIBRONEER-IPF is one record with NCT05321069 and EudraCT 2022-001091-34.
The source on every fact
Registry ids, DrugBank ids and guideline identifiers travel with each record, so you can open the original.
What was checked, and what was left
Each screen states its evidence, its confidence and what it left unexamined. A green light is a reason to investigate; the judgement stays with you.
Public sources
The twin is built from trial registries, the biomedical literature, drug and regulatory records and patents. Your workspace holds what you add to it, visible to you.
Your first steps
From registration to your first question.
- 1
Register
Two minutes, from public information: your name, your institution, your research in a line.
- 2
Your research profile
Drafted from your trials and publications. You check and correct it.
- 3
Connect your assistant
Add MondoTwin as a connector in Claude, ChatGPT or any assistant that supports MCP, and sign in once. The address comes with your approval email.
- 4
A matching mandate
When a pharma mandate fits your work, we contact you before anything is shared.
Maestro
Great science lives twice.
Once when it is published. Again when it reaches a patient. Somewhere a family counts the wait in birthdays; somewhere a researcher holds the idea that could shorten it. The twin reads existing science again, as a pianist reads a score, and carries it to the companies able to develop it. The signature stays with its author.
- Authored.Every discovery has a person behind it. The name stays on the work.
- Read again.Existing science, reinterpreted against the whole of the evidence.
- Delivered.Carried to the companies able to develop it, and on to the people waiting.
Your terms
Free of charge, in your name.
Pharmaceutical companies pay for the department; researchers and institutions use the platform free of charge. You start with public information, and confidential material follows an agreement with your institution. Your data, your authorship and your institution’s rights stay where they are. A partner sees what you approve, and one email removes your registration. The evidence tools are for academic use, with a university email; access is approved per registration, and one email removes it.
