Connector tools
When you connect Gosset to ChatGPT, Claude or your own agent, the assistant gets the tools below. You don't call them yourself: ask a question and the assistant picks the tool and its filters. This page tells you what you can ask for. If you build your own agent, the names are the tool names your agent sees.
Some tools are available only on plans that include them. Your assistant lists the tools your account can use.
Search
| Tool | What it finds |
|---|---|
find_drugs | Drugs and candidates by target, disease, modality, company or phase |
find_trials | Clinical trials, with design, endpoints, population and results summaries |
find_companies | Company profiles and pipelines |
find_deals | Licensing, partnership and M&A deals |
find_news | News, press releases and filings |
find_events | Dated company and conference events |
find_trial_results | Structured trial outcomes, with the paper or registry values |
find_conference_abstracts Added 17 November 2026. Search conference abstracts across meetings from your assistant. | Conference abstracts across meetings, with linked trials and impact scores |
Search results include a view_url Added 23 October 2026. Search results link to the same search as a table in the web app. that opens the same search as a table in the Gosset web app.
Explorers and charts
| Tool | What it shows |
|---|---|
show_drug_landscape | A landscape of drugs for a target or disease |
show_trial_explorer, show_deal_explorer | Interactive tables of trials or deals. limit sets how many rows (up to 200, and up to your plan's cap) Added 15 October 2026. Choose how many rows the trial and deal explorers show. |
show_result_explorer | Charts of reported results |
show_conference_explorer Added 17 October 2026. Browse conference presentations by meeting, year, drug, target or company. | Conference presentations by meeting, year, disease, target, drug, company or keyword, sorted by date or impact |
get_drug_evidence | The evidence behind one drug's profile |
Predictions and reports
| Tool | What it does |
|---|---|
estimate_ptrs, predict_trial_success | Probability of technical and regulatory success |
estimate_remaining_time | Expected time to trial completion |
report_submit, report_fetch, report_list | Longer reports that run in the background. email_when_ready sends a results comparison to your email when it is ready Added 17 October 2026. Get an email when a results comparison report is ready. |
Drugs: what you can filter and ask for
find_drugs filters by target, disease, modality, company and phase, and also by:
| Filter | Values |
|---|---|
adc_payload, adc_linker, payload_mechanism_targets Added 12 November 2026. Ask your assistant for ADCs by payload, linker or payload target. | ADC payload class (such as topo-I inhibitor), linker (cleavable, non-cleavable or a chemistry), and what the payload acts on (such as TOP1) |
vaccine_valency, vaccine_pathogen Added 12 November 2026. Ask your assistant for vaccines by pathogen or valency. | Valency word (monovalent … nonavalent, multivalent) and pathogen name |
accelerated_approval_status, accelerated_approval_overdue Added 12 November 2026. Ask your assistant for FDA accelerated approvals by status or overdue requirements. | ongoing, verified, withdrawn; and whether FDA's due date has passed |
Ask for more detail on a drug and the assistant can return ADC detail, vaccine composition (including how many pathogens a vaccine covers), FDA approvals per indication, approvals by other regulators with links, and FDA accelerated approvals.
Approval numbers and approval status by regulator, combination_vaccine and the preclinical stage are in the CLI and the REST API, not in the connector yet.
Trials: population and design
find_trials filters by drug, target, disease, sponsor, phase and status, and also by the population the eligibility criteria describe (beta):
| Filter | Values |
|---|---|
max_ecog_min, max_ecog_max Added 6 November 2026. Ask for trials by the highest ECOG score they admit. | Highest ECOG score the trial admits, 0–5 |
prior_lines_min, prior_lines_max Added 6 November 2026. Ask for trials by the number of prior lines they require. | Prior lines of therapy the trial requires, 0–20 |
treatment_naive Added 12 November 2026. Ask for trials in treatment-naive patients. | true or false |
stage_group Added 14 November 2026. Ask for trials by disease stage at entry. | early, locally_advanced, advanced, metastatic |
severity Added 14 November 2026. Ask for trials by the severity they require. | mild, moderate, severe |
treatment_setting Added 14 November 2026. Ask for trials by treatment setting, such as adjuvant or first line. | neoadjuvant, adjuvant, perioperative, first_line, second_line, third_line_or_later, previously_treated, maintenance, induction, consolidation |
Each trial carries population_context Added 14 November 2026. Trials show their population at entry, with the eligibility text behind it.: stage, severity, treatment setting, treatment-naive status and the therapy classes patients must have had before entry, each with the eligibility text it rests on. Ask for composite primary endpoints, such as MACE, and the assistant returns their families and components Added 6 November 2026. Ask for trials with composite primary endpoints, such as MACE..
News and deals
find_newsfilters bysource_typeAdded 1 November 2026. Ask for news from one kind of source, such as press releases or SEC filings.:press_release,news,sec_filing,hkex_filing,wechat,other. Titles are in English, with the original title and its language alongside.find_dealsreturnsprovenance_statusAdded 8 November 2026. Deals show whether a primary document backs them. (verified,reported,unverified,not_a_transaction) and can filter on it. Records that are not transactions are left out unless you ask for them. A company you name is matched on either side of a deal.
Conference abstracts
find_conference_abstracts searches presentations by conference and year, drug, company, disease, target, modality or linked trial, with free-text search. Sort by date or by Gosset's impact score (0–10). Ask for the abstract text when you need it.
Try the research examples for questions that use these tools.