AI NewsProductivityReported
ActivTrak says heavy AI users work 90 minutes longer per day
ActivTrak measured 9,140 knowledge workers at 467 companies. Heavy AI users worked up to 90 minutes longer per day than non-users.
Why it mattersThe sales pitch is that AI tools save time. In this measured set, the heaviest users ended up with the longest days and the shortest stretches of focus.
The pitch for a coding assistant or an in-app AI tool is that it saves the user time. A new report from the ActivTrak Productivity Lab looks at what people actually did with that time. In its measured set, heavy AI users worked longer days, not shorter ones.
ActivTrak says it watched the digital work patterns of 9,140 knowledge workers across 467 companies. The window ran from 1 April to 25 September 2026. The company sells software that tracks what employees do on their screens, so it has an interest in the result. The numbers are its own.
What the report says
Workday length is the headline number. ActivTrak says a heavy AI user put in a workday of 8.4 to 9.1 hours. A non-user put in 7.6 hours. The gap is up to 90 minutes, or one extra working hour and a half.
The focus numbers go the other way. ActivTrak says an uninterrupted focus session averaged 37.8 minutes for a heavy AI user. For a light user, the average was 48.2 minutes. That is 22 percent shorter.
Heavy users also switched between applications more often. ActivTrak says they averaged 1.1 switches per hour, against 0.95 for non-users. Formal meetings were 3 percent of a heavy user's time, against 5 percent for a non-user.
One more number worth knowing
ActivTrak noted a split by how the AI is delivered. A tool built into the application the person is already using led to fewer application switches. A tool opened in a browser led to more. The report does not give a single number for the split.
What this does not prove
The study is a correlation. People who already work longer days may be the ones who adopt AI tools first. A heavy user of a coding assistant may also be the person who takes on the harder tickets. ActivTrak does not say its sample is a random draw from all knowledge workers. The companies that pay for an activity tracker are not a random sample of all companies either.
What to do with it
A team that pays for Claude Code, Cursor or Copilot may have expected a shorter day. This is a reason to measure. Pull the time logs. Compare the people using the tool heavily with the ones who are not. The tool is sold on time saved. That promise has to show up in time logged, not in stories about speed.
Source
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.
