ActivTrak's latest workplace analysis associates AI use with longer workday spans. It does not establish that adopting AI causes an employee to perform 90 more minutes of work each day. That distinction matters when deciding whether an AI rollout is saving time.
The research release is dated October 29, 2026; Gizmodo's originating coverage appeared on November 3. The observations cover April 1 to September 25: 9,140 knowledge workers across 467 enterprises.
Read the time measures separately
ActivTrak reports workday spans of roughly 8.4–9.1 hours among AI users, against 7.6 hours among non-users. A span describes the extent of a day; it should not be read as continuous active work or completed output.
The company's detailed analysis also separates focus sessions from after-hours activity. Heavy users averaged 37.8 minutes per uninterrupted focus session, versus 48.2 for light users. Heavy users are defined as exceeding 45 minutes of AI use and having at least five sessions daily; light users use it for less than ten minutes.
Those comparisons answer different questions. A shorter focus session is not itself proof of poorer work, just as a longer workday does not reveal how much useful work was delivered.
The workflow is a plausible explanation to test
The analysis associates embedded AI with fewer application switches than browser-based AI. It also describes differences between engineers and operations managers. These observations make integration worth investigating, but they do not isolate the effect of an interface from the job performed.
This is research produced by a workplace-analytics vendor using observed behavior. The published material does not present randomized assignment that would separate AI's effect from differences in roles, workload or adoption choices. The headline gap should therefore prompt questions about work design rather than a causal verdict about AI.
Give a pilot an outcome and a boundary
TechKili's assessment: define a completed task before measuring a pilot, then compare time, rework and quality for comparable assignments. Ask employees whether the saved time removes work or merely makes room for more requests. Agree on working-hour boundaries and explain what activity data will be collected.
A tool-usage target can be met while the broader process remains slow. The practical test is whether a team can deliver the intended result with less friction and a sustainable workload.
Sources
Related Articles
PsiQuantum samples an 8×8 optical switch for AI-network evaluation
The quantum-computing photonics platform has a networking demonstrator. Its lab figures explain the component, while deployment still requires system evidence.
Artificial Analysis v5 changes the yardstick for choosing AI models
The updated Intelligence Index changes its tasks and weights. Teams choosing a model need version-aware comparisons, matching effort settings and their own cost evidence.
Anthropic's IPO filing ties much of its $80.7 billion loss to valuation changes
The public S-1 reports a much smaller operating loss and positive operating cash flow for the same six months. These measures explain different parts of the business.
Apollo's AI payoff claim hinges on how productivity is measured
Torsten Slok's October 24 note questions broad AI returns. Its index comparisons and productivity charts are useful context, but they do not measure a particular company's deployment.