What it is

Microsoft, which sells Copilot and agent tooling. Survey fieldwork by Edelman Data x Intelligence.

Fielded February to March 2025, with 1,000 workers per market and an additional US sample. 82% of leaders say they will use digital labor to expand capacity in the next 12 to 18 months, and 45% call it a top priority, while 80% of workers say they lack the time or energy to do their jobs. Leaders and employees disagree on readiness: 67% of leaders are familiar with agents against 40% of employees, and 36% of leaders expect to manage agents against 21% of employees.

The report defines a “Frontier Firm” by organization-wide deployment, active agent use and belief that agents drive return, and describes a three-phase path from assistant to digital colleague to agents running whole workflows with people setting direction. Leaders in those firms use AI for marketing tasks at 73% against 55% globally.

What marketing teams should take from it

  • Its plans for agent adoption and new roles are expectations. Chapter 4’s 46% is what marketing teams report has actually changed.
  • The leader-employee gap on agent familiarity is the same shape as the readiness gap in finding 04.4, measured across a much larger sample.
  • The human-to-agent ratio for a workflow is a planning variable worth writing down for one task.
  • Marketing appears as one task category among many; the sample is knowledge workers broadly.

Where to start

Ask your functional heads and ten of their people the same question, whether they could describe what an AI agent does, and compare the split with 67% and 40%. Then decide a human-to-agent ratio for one workflow such as first-draft content.

Use the 45% of leaders who plan to keep headcount flat and add digital labor, a separate figure from the 45% who call capacity a top priority, as the explicit alternative to headcount cuts in the next planning round.

Keep in mind

Vendor-authored and self-reported intent, fielded before most agent tooling matured. Outcome comparisons pit a selected leader group against all workers.

Where the report uses it

Read during the study design; not cited in the report or the perspectives.