Creating and editing documents is among the most frequently reported uses of artificial intelligence at work, according to research from Adobe that examines how employees use software to draft, summarise and manage information.
The Rise of AI Documents study draws on a survey of 2,000 UK adults conducted by 3Gem for Adobe Acrobat between 18 and 22 April 2026. Its findings were circulated in October, but the fieldwork was carried out in April and should not be presented as a fresh October survey.
Within the survey’s reported workplace AI use categories, 27% identified creating and editing documents, including emails, reports, proofreading and summarising PDFs. Another 23% selected data analysis or technical tasks, while 22% identified assistance with managing tasks such as notes, scheduling and brainstorming.
These percentages describe responses to questions about use cases, not controlled observations of how much time each task takes. They also do not establish that more than a quarter of the entire UK workforce uses a particular Adobe product.
The survey recorded marketing and sales activity among reported AI applications at 16%, while translating or localising content and human resources or operations tasks each appeared at 14%. The differences illustrate the range of activities employees associate with AI rather than the adoption level of any single software platform.
Adobe also reported considerable variation between industries. Among respondents in fast-moving consumer goods roles, 50% cited creating or editing documents as a use for AI. In IT, technology and telecommunications, 47% identified data analysis or technical work.
The company found that 44% of respondents working in marketing used AI for meeting and task management. Such figures relate to occupational subsets within the survey and should not be presented as though every industry group contained the same number of respondents.
The research describes perceived time savings from using AI with workplace information, but these benefits are based on participant accounts rather than independent time-and-motion testing. A person who says a tool saves time may still need to review accuracy, formatting and compliance before a document can be used.
That distinction matters for employers comparing productivity claims. Generating a first draft can reduce the initial writing effort, while verification and approval may require additional work. The net result depends on document complexity, employee skill and the reliability of information supplied to the software.
The survey also identified barriers to adoption. Adobe reported that 22% of respondents mentioned data security and privacy concerns, while 20% were worried about inaccurate outputs. A further 17% preferred human input over AI-generated content.
For 19%, the challenge was insufficient training or guidance. Another 14% referred to company policies restricting AI use, and 12% reported uncertainty about which tools to choose. These responses suggest that organisational processes can affect adoption alongside the technical capability of the tools themselves.
Some document tasks carry particular business risks. Contracts, personnel records, customer correspondence and financial reporting may contain information that should not be copied into an external service without suitable permissions. Companies need to establish acceptable use, data retention and review responsibilities before deploying automation widely.
Adobe supplemented the survey with search data collected through DataForSEO. Its largest task-based category, work writing and administration support, generated 122,010 recorded searches under the study’s grouping methodology. Editing and formatting documents ranked next with 73,150 searches.
The search figures should not be interpreted as numbers of unique people using AI or a count of completed workplace tasks. The company grouped related search terms into task categories and examined search volume, which measures interest rather than confirmed adoption or successful outcomes.
Adobe also reported regional changes in those searches, including increased interest in work writing and administration in Bradford, document editing in Southampton and translation or localisation in Belfast. Such changes describe the selected search series and do not independently establish productivity differences between cities.
The company’s methodology notes that a clean division between AI and non-AI searches was not feasible, because apparently general terms such as translating or editing a PDF may lead to AI-assisted tools. The final analysis therefore grouped practical tasks rather than claiming every search explicitly requested an AI product.
For businesses considering these systems, the research points to document preparation, information retrieval and routine coordination as common areas of interest. The commercial case will depend on measured outcomes after training, verification and governance costs are included, rather than on survey responses or search demand alone.
Adobe’s April survey provides a snapshot of reported workplace habits and concerns, not proof that the same level of adoption or time saving applies to every UK employer in October. Its value lies in identifying the tasks people are trying to improve and the conditions that may limit the benefits.
You must be logged in to post a comment.