Skip to main content

AI Usage > Created at

Written by Tom Williams

Dataset: AI Usage

Entity: AI Usage Measurement

Field ID: created_at

Type: Datetime

Description: The UTC midnight of the day the measurement covers. AI usage is reported per calendar day, so this is the day the measured activity belongs to rather than a record creation timestamp.

Source: App

Transformation logic: The provider's day or time bucket is normalised to UTC midnight.

App Mapping

GitHub Copilot

day at 00:00:00Z

Anthropic Claude (coming soon)

the report date at 00:00:00Z

Cursor (coming soon)

date at 00:00:00Z

OpenAI Platform (coming soon)

the usage bucket's start_time, truncated to the UTC day

OpenAI Codex (coming soon)

the usage bucket's day at 00:00:00Z

Reporting Use Cases

The Created at field is the time dimension for every AI usage report. Because each record covers a whole UTC day, it behaves slightly differently from the same field on other datasets.

  • Adoption over time: Group by YEAR_MONTH(created_at) or YEAR_WEEK(created_at) with a measure such as SUM(interactions) to track how AI usage is growing.

  • Active developers per period: Combine with COUNT_DISTINCT(actor_username) and aggregation_level = USER to count how many people used an AI tool in a period.

  • Do not group finer than a day: Every record sits at midnight, so grouping by hour produces a single bucket.

  • Use UTC: Providers bucket their own days in UTC. Reporting in another timezone shifts part of each provider day into the neighbouring day and makes daily figures look uneven.

  • Expect recent days to move: Late-arriving telemetry means the last few days of a trend will keep climbing for a while after the fact.

Did this answer your question?