For the past three years, most enterprise use of generative AI has followed a familiar pattern: an employee asks a question, the model answers, and the employee does the rest. That pattern delivered real productivity gains, but it left the hardest part of most jobs, coordinating information and actions across many systems and people, with the human. The next phase moves the unit of work from the prompt to the outcome, as employees assign an objective to an agent that plans the task, uses the necessary tools and data, and returns the finished work.

Organizations are ready for that shift. According to Futurum’s 1H 2026 AI Decision Maker Survey, 64% are already piloting or deploying agentic AI, yet only 11% describe their AI maturity as transformational. First-generation agent programs are running into limits on governance, integration, cost, and model churn. Moving from assistance to delegation requires an agent platform with persistent context, presence where work happens, orchestration, model choice, and governance and cost control designed in from the start.

In our latest thought leadership brief, Delegating Outcomes: How a Universal Agent Changes the Way Enterprises Work, completed in partnership with Google Cloud, Futurum Research examines why first-generation agent programs hit a ceiling, defines five requirements for a universal agent for work, and assesses how Google’s Gemini agent addresses them.

In this brief, you will learn:

• Why 64% of organizations are piloting or deploying agentic AI, but only 11% report transformational AI maturity
• The five requirements that separate a universal agent for work from a collection of assistants
• What delegation changes for CxOs, IT decision makers, knowledge workers, and developers
• How Google’s Gemini agent approaches the universal agent, with early results from Commerzbank and Nokia

If you are interested in learning more, be sure to download your copy of Delegating Outcomes: How a Universal Agent Changes the Way Enterprises Work today.