How Precedia works

Precedia is built from public regulatory documents only. Every answer is a verbatim quote with its source and page, checked by machine before it is published.

What Precedia covers

453 products approved by FDA or through the EU centralised procedure, across eight modalities: monoclonal antibodies (117), multispecific antibodies (18), antibody-drug conjugates (14), oligonucleotides (27), gene therapy (26), cell therapy (27), small molecules (211), protein and vlp vaccines (13). Coverage aims to include every approval in these modalities as of 13 Oct 2026. Radioconjugates, immunotoxins and products approved only outside the US and EU are out of scope. A product is left out when no review documents are public.

Sources

• FDA review documents (multi-discipline, pharmacology/toxicology and integrated reviews) from Drugs@FDA.
• FDA CBER approval packages for gene and cell therapies: summary basis for regulatory action and pharmacology/toxicology review memos.
• EMA European public assessment reports (non-clinical sections).
• US prescribing information (nonclinical toxicology sections).

Extraction

For each product, AI agents extract the same 15 fields (species and rationale, surrogates, study length and design, genotoxicity, carcinogenicity, reproductive toxicity, safety pharmacology, first-in-human dose, NOAEL, immune findings, biodistribution and modality-specific risks). Each value must come with a verbatim quote, the document and the page.

Verification

Every quote is checked, word for word, against the text of the page it cites. If one quote fails, the build fails and nothing is published. Today 13,619 of 13,619 quotes pass.

How Precedia stays current

A scheduled job checks the sources every week: new FDA approvals and review documents (Drugs@FDA and CBER), new EMA authorisations and assessment report updates, and changes to ICH and FDA guidance pages. Anything new goes into a review queue. Extraction and verification run on everything in the queue, and an update is published on the first Monday of each month.

• Data updated: 13 Oct 2026
• Next update: 7 Dec 2026
• Sources last checked: 2 Nov 2026

Summaries and counts

Board dots and counts (species, conducted or skipped studies) are derived from the extracted text and spot-checked. How the first human dose was set is classified by hand: 10 of 453 reviews state it; the rest are marked partly stated or not stated. Study lengths worked out from a dosing schedule are labelled “derived”.

Limits

• Public documents only. Much of the first-in-human rationale sits in confidential IND submissions, so it is often not stated publicly.
• Page numbers refer to the page of the PDF file, which can differ from the number printed on the page.
• Precedia is educational and is not regulatory advice. Confirm decisions with your regulatory team and, where appropriate, with the agency.