Palanor holds the documents behind its scores: S&P 500 earnings call transcripts, earnings releases, 10-K risk factors and 10-Q MD&A, plus the Federal Reserve's Beige Book, FOMC statements and minutes, and Fed speeches. You can search all of them by meaning, not only by exact words.
Asking Numen
Ask Numen in plain words. "What did Microsoft say about AI capex last quarter?" or "What did the latest Beige Book say about hiring?" Numen searches the documents, reads the passages that match, and answers from them. Every claim it takes from a document links to the passage, so you can read the words yourself. If no passage answers the question, Numen says so rather than filling the gap.
Searching one company
On a company's Discourse page, the Search the documents block takes a phrase, such as "pricing pressure" or "data center capacity". Results are passages in the company's own words, newest and closest first, with no more than two from any one document. Filter by calls, releases, 10-Q MD&A or 10-K risk factors. The block also shows how many of each document we hold for the company.
Opening a passage
Every result opens the document reader at the passage, shaded so it stands out from the text around it. The reader is described in [Reading the documents behind a score](/help/modules/reading-the-documents-behind-a-score).
Palanor summaries
A call, release, filing or Fed document with a Palanor summary shows it at the top of the reader, marked Palanor summary. It is our one-paragraph reading of the whole document, written by Palanor. It is not the company's words, and never a quote. When the exact words matter, read the passage.
What you can see
Call transcripts are licensed to Palanor. You see them, and Numen searches them, only while signed in. SEC filings and Federal Reserve documents are public records.
In the API and MCP
/v1/documents/search in the API and palanor_search_documents in the Palanor MCP search SEC filings and Federal Reserve documents only. Each result carries the passage, the company, the date, a link to the public reader on palanor.com, and its source. Call transcripts are never available there.
How search works
Each document is cut into passages of a few paragraphs, about 450 words, along the reader's own paragraphs and speaker turns. A call's passages never mix prepared remarks with the question-and-answer session. Each passage is stored with the company, document type and date, so a search can be narrowed to one company, a date range or a kind of document, then ranked by how close its meaning is to what you asked. New documents are added within the hour.
Related: [Reading the documents behind a score](/help/modules/reading-the-documents-behind-a-score) · [Meet Numen](/help/numen/meet-numen)