Dataset: AI Usage
Entity: AI Usage Measurement
Field ID: requests
Type: Number
Description: The number of model or API calls, including automated follow-ups. Not interchangeable with interactions or sessions: on agentic surfaces this is far larger than the number of human prompts. Most providers publish only some of the three, so an empty value is missing coverage rather than a zero.
Source: App
Transformation logic: N/A
App Mapping
GitHub Copilot |
|
Anthropic Claude (coming soon) | N/A - neither endpoint publishes a request count |
Cursor (coming soon) | the sum of the subscription, API key and usage-based request counters on day total records; the number of usage events on per-model records |
OpenAI Platform (coming soon) |
|
OpenAI Codex (coming soon) | N/A |
Reporting Use Cases
The Requests field counts calls to the model, including the automated follow-ups an agent makes on its own. It measures machine work, where interactions measures human work.
Cost and capacity planning: Requests track consumption far better than interactions once agents are in use, because an agent turns one prompt into many calls.
Agentic leverage: The ratio of requests to interactions, over records that carry both, shows how much work each prompt sets in motion.
Not a measure of adoption: A rising request count with flat interactions means agents are doing more per prompt, not that more people are using AI.
Empty is not zero: Anthropic and Codex publish no request count at all.