Skip to main content
The Document Corpus holds the earnings call transcripts, conference transcripts (broker and investor conferences, analyst and investor days, keynotes, deal calls and shareholder meetings), SEC filings and company documents that semantic retrieval searches. Three endpoints address it directly: list a company’s documents, fetch one document’s metadata by id, and read a document’s ordered text segments. Each document is associated with a ticker, a year and quarter where it has one, and, for filings, a source URL.
Plan: Free and above · Credits: 1 per call · MCP tools: list_documents, get_document, get_document_segments

​
What You Can Use the Document Corpus For

Discover which periods exist for a company before searching one. Read a full document in order rather than as ranked fragments. Resolve a document id returned by retrieval back to its metadata. List the conferences and investor days a company’s management appeared at, by date.

​
Why to List Documents Before Searching

The document listing is the authoritative answer to which fiscal periods a company has in the corpus. Searching a guessed period returns stale passages rather than an error, so listing first is how a wrong period is caught before it becomes a wrong answer.

​
Document Corpus Date Convention

year and quarter are the fiscal period the document covers, as the company labels it, not the calendar date it was published. A conference transcript is the exception: its year and quarter are the calendar quarter in which the event took place. Documents are listed most recent first.

​
Key Document Corpus Fields

document_id identifies a document and is the key the fetch and segments endpoints take. ticker identifies the company. year and quarter place it in fiscal time. The document type distinguishes an earnings call (earnings_call) from a conference transcript (conference_transcript) and from a filing. source_url links the original filing; it is null on every transcript, earnings call or conference. A segment is a contiguous ordered passage of a document. Segments are the unit retrieval ranks and cites, and reading them in order reconstructs the document.

​
Document Corpus Sources

Transcripts are sourced from the calls and events themselves; filings come from SEC EDGAR. Filings keep their source_url so any passage can be traced to the original; transcripts are served without a source link, and are traced by their document_id, title and date.

​
List a Company’s Documents

Use the documents-by-ticker endpoint, which returns that company’s documents most recent first.
​
string
required
The company’s ticker.
​
integer
Restrict to one fiscal year, between 2000 and 2100.
​
string
Restrict to one fiscal quarter: Q1, Q2, Q3 or Q4.
​
integer
default:"20"
Maximum documents to return, up to 100.
​
integer
default:"0"
Pagination offset. This endpoint pages by offset, not by cursor.

​
Fetch One Document

Use the document endpoint with a document_id to retrieve that document’s metadata.
​
string
required
The document_id, as returned by the listing endpoint or by retrieval.

​
Read a Document’s Segments

Use the document segments endpoint to read a document’s ordered text.
​
string
required
The document_id.
​
integer
default:"20"
Maximum segments to return, up to 100.
See also Semantic Retrieval to search the corpus by meaning, Earnings Call Transcripts for what the corpus holds, and SEC Filing Items for a named section of a named filing.