Large-Cap AI Mentions per Call (Discourse)

## What this measures

How often management substantively mentions AI, agents or large language models, averaged per document, across member companies in the trailing 90 days.

## Why it matters

Management teams speak for their businesses every quarter. Read across large-cap companies, their language shows what this group of companies expects before it appears in results or official data.

## How to read it

Rising means AI is taking up more of what management chooses to talk about.

## What it does not say

Talk is not spending or results; a mention can be a plan, a product or a risk. Readings before 9 October 2026 use each company's tier on that date, applied backward, so a company that has grown or shrunk across a boundary is read in its current tier throughout.

## The universe

Large-cap companies: those worth $10 billion to $200 billion, the bulk of the S&P 500. Market value is SEC-filed shares outstanding times the latest close, updated daily. A company changes tier only once its value is 10% past the boundary (for example above $11 billion to move up into large cap, below $9 billion to fall back to mid cap), and the change applies from that date. Companies Discourse tracks are the members of the S&P 500, Nasdaq-100, Russell 1000 and S&P MidCap 400; companies without a filed share count (most foreign issuers) have no tier. A value is published only when at least 70% of the companies in the tier on that date, and at least 300, have a scored document in the 90-day window; on dates below that, nothing is published.

## Source

Palanor Discourse, derived from earnings-call transcripts received through a licensed data provider (press releases filed with the SEC where no transcript exists). One document per company per calendar quarter, trailing 90 days. Each document is scored by the same model and prompt as the S&P 500 Discourse family. Scored by Palanor.