Dropbox survey: AI users spend 36.6% of their AI time hunting for context

AI is a large part of the workweek

Dropbox surveyed 219 professionals, primarily working in technology, AI, software, finance, and healthcare, at The AI Conference 2026 in San Francisco. The report brings together 70+ statistics on how AI is changing content at work.

According to Dropbox, respondents spend an average of 15.8 hours per week actively engaging with AI at work. That figure excludes auto-running Codex jobs that operate on autopilot. Only 2.3% of respondents report zero hours with AI in an average week.

• 81.3% spend at least six hours a week working with AI.
• 48.8% spend more than 15 hours.
• 24.7% report 6 to 10 hours, the largest group.
• 21.0% report 30 or more hours.

Context retrieval eats into productivity

Respondents estimate that 36.6% of their AI engagement time goes to finding, opening, uploading, or cross-referencing documents, messages, links, and information in other work apps. Applied to a 15.8-hour AI workweek, Dropbox calculates this at approximately 5.8 hours a week. Nearly one in three respondents spend at least half of their AI time retrieving context, while 1.8% report spending none.

The effort continues after the first prompt. Respondents report an average of 4.3 additional context rounds before a result is ready to use. Dropbox says 37.4% add context five or more times, and 10.0% add it at least 10 times. The company also warns about quality: if AI lacks relevant team and business information, users must spot the gaps themselves, and a missed decision, requirement, or source document can leave a result incomplete or inaccurate and lead to rework.

Dropbox also puts a cost on this. Applying the survey averages to a 100-person company, it suggests approximately 27,691 hours a year spent retrieving context, valued at $1.15 million in salaries alone. Cutting that time by 50% through connected context and content could recover approximately 13,846 hours annually, or $576,952 in potential productivity savings. For a 1,000-person company, Dropbox puts the opportunity at $5.77 million a year. These are the company's own estimates built on survey averages.

More files, and much of the useful output stays in chats

The report also looks at content creation. Respondents estimate that 46.4% of the new work files they create or save are created or substantially changed with AI. The key findings list a 162.6% average growth in new work files created since AI adoption, and 93.2% of people say they create more work files now with AI.

Dropbox says useful output is spreading across shared workspaces, local files, and chats, leaving teams with a growing body of knowledge to organize, share, and put to work. According to the report, 43.8% of useful results created by AI never leave the prompt.

What it means for businesses

Dropbox draws a practical conclusion from its data: making relevant content easier to find and bring into the AI workflow is an opportunity to improve productivity. In its view, less time spent gathering context would give employees more time to develop useful results and move projects forward. The findings come from a self-reported survey of a specific conference audience, so they describe how these professionals estimate their own habits rather than the workforce as a whole.