DealNode

How it works

From newswire to
defensible answer.

DealNode reads the life-sciences newswire and turns it into a connected, queryable map of companies, investors, trials, targets, and partnerships, so the questions that used to take a week of manual research become a single query. Here is how the map gets built: four stages, in order, every day, with an evidence record on every relationship it writes.

01 / Ingest

Read the sources your analysts already read

Trade press, ClinicalTrials.gov, regulatory filings, and company releases, monitored continuously. New financings, readouts, and licensing deals enter the pipeline the day they are published.

02 / Extract

Turn news into structure, not summaries

Language models pull entities, events, and relationships against a strict life-sciences ontology, never free-form. Each extraction records its source snippet and model version so nothing is unattributable.

03 / Resolve

One company, not four spellings

Tiered entity resolution merges duplicates against external identifiers, so “Arch”, “ARCH Venture”, and “Arch Venture Partners” are one node. Low-confidence writes queue for human review before they become fact.

04 / Serve

Query, watch, or ask, with citations

Screen the map by indication, mechanism, phase, and partnership status; subscribe to alerts on the entities you track; or ask a plain-language question and get an answer computed over cited relationships.

The standard

Intelligence you can't cite is opinion.

Every relationship in the map stores its source URL, a verbatim snippet, and its extraction record. That is what lets you put an answer in front of an investment committee or a board without caveats. It is why we say, plainly, that if we can't source it, we don't state it. The figures shown across this site are illustrative placeholders, labelled as such; the real map is built only from cited events.

See a sample map and dossier →

See your corner of the map.

Early access is open for life-sciences investors, BD teams, and founders.