By Jennifer Zou, Profound
A new cross-platform analysis from Profound finds that AI assistant use follows a pronounced daily and weekly rhythm, and that the rhythm differs by platform, region, and demographic group.
The study examined roughly 300,000 conversations from users across 27 countries, drawn from ChatGPT, Claude, and Gemini over a one year period between July 2025 and June 2026. Most large usage studies to date have covered a single product or geography, making this the first to compare global temporal patterns across the leading assistants at once.
The platforms are used for different purposes. Work accounts for 71% of Claude conversations, compared with roughly 43% on Gemini and 40% on ChatGPT. ChatGPT also shows the sharpest intra-day swing, a 3.6 percentage point daily spread from its work-share peak at 10am to its trough at 9pm local time. Claude's spread is about half as wide, at 2.11 points.
Work-related prompts climb from an overnight floor near 2% of daily volume to a peak around 11am local time, then fade through the evening, while personal use runs flatter and takes over at night. The vast majority (between 76% and 78%) of conversations on all three platforms occur on weekdays.
Geography drives much of the pattern. Europe shows the sharpest separation between professional and personal use through the afternoon, while in North America the two track each other almost all day. Every region except North America shows a lunchtime dip in volume around noon, most pronounced in Asia, where it persists into weekends.
Age and income sort users into schedules. Users aged 30 to 49 form the densest weekday-midday block, those 65 and older concentrate in weekday late mornings, and the youngest cohort (under 30s) peak on Sunday afternoons and evenings. Households earning $100,000 to $200,000 show the tightest weekday work-hours block, while those under $25,000 are spread across nearly all hours and every day of the week.
Each assistant has a topic specialty. On weekdays, writing accounts for about 28% of ChatGPT's conversations, programming about 35% of Claude's, and multimedia generation about 29% of Gemini's. Weekends shift the mix as ChatGPT's writing share falls to 21%, while Claude's programming share holds level.
Methodology. Conversations were classified by an LLM-based judge using the topic taxonomy from OpenAI's How People Use ChatGPT study, validated against the public WildChat dataset. Prompts were anonymized and summarized before analysis, and no human read a raw user query. The full analysis includes figures, confidence intervals, and complete methodology.
AI disclosure: AI tools were used for editing and proofreading only. All research, analysis, and writing were performed by the author.
Reviewed by Irfan Ahmad.
Read next: Americans’ Social Media App Usage Is Up. Its Share of Screen Is Not
A new cross-platform analysis from Profound finds that AI assistant use follows a pronounced daily and weekly rhythm, and that the rhythm differs by platform, region, and demographic group.
The study examined roughly 300,000 conversations from users across 27 countries, drawn from ChatGPT, Claude, and Gemini over a one year period between July 2025 and June 2026. Most large usage studies to date have covered a single product or geography, making this the first to compare global temporal patterns across the leading assistants at once.
The platforms are used for different purposes. Work accounts for 71% of Claude conversations, compared with roughly 43% on Gemini and 40% on ChatGPT. ChatGPT also shows the sharpest intra-day swing, a 3.6 percentage point daily spread from its work-share peak at 10am to its trough at 9pm local time. Claude's spread is about half as wide, at 2.11 points.
Work-related prompts climb from an overnight floor near 2% of daily volume to a peak around 11am local time, then fade through the evening, while personal use runs flatter and takes over at night. The vast majority (between 76% and 78%) of conversations on all three platforms occur on weekdays.
Geography drives much of the pattern. Europe shows the sharpest separation between professional and personal use through the afternoon, while in North America the two track each other almost all day. Every region except North America shows a lunchtime dip in volume around noon, most pronounced in Asia, where it persists into weekends.
Age and income sort users into schedules. Users aged 30 to 49 form the densest weekday-midday block, those 65 and older concentrate in weekday late mornings, and the youngest cohort (under 30s) peak on Sunday afternoons and evenings. Households earning $100,000 to $200,000 show the tightest weekday work-hours block, while those under $25,000 are spread across nearly all hours and every day of the week.
Each assistant has a topic specialty. On weekdays, writing accounts for about 28% of ChatGPT's conversations, programming about 35% of Claude's, and multimedia generation about 29% of Gemini's. Weekends shift the mix as ChatGPT's writing share falls to 21%, while Claude's programming share holds level.
Methodology. Conversations were classified by an LLM-based judge using the topic taxonomy from OpenAI's How People Use ChatGPT study, validated against the public WildChat dataset. Prompts were anonymized and summarized before analysis, and no human read a raw user query. The full analysis includes figures, confidence intervals, and complete methodology.
AI disclosure: AI tools were used for editing and proofreading only. All research, analysis, and writing were performed by the author.
Reviewed by Irfan Ahmad.
Read next: Americans’ Social Media App Usage Is Up. Its Share of Screen Is Not
