Microsoft’s Work Trend Index, published on 23 April 2025, describes a business model it calls the “Frontier Firm”: an organisation combining human judgement with AI agents and on-demand intelligence. The report analysed survey data from 31,000 workers across 31 countries alongside other research inputs.
Among the surveyed leaders, 81% expected agents to be moderately or extensively integrated into their company’s AI strategy over the following 12 to 18 months. That is an expectation reported in 2025, not a measurement of completed deployment today.
From assistance to delegated work
The report sets out a progression from AI assistants to agents handling specific tasks and eventually wider workflows under human direction. It also says the progression is not necessarily linear: different parts of an organisation may be at different stages.
For a business software buyer, the terminology can be less important than the operational detail. An agent that suggests an answer carries different risks from one that can change a customer record, approve a payment or send a message.
Control before scale
Analysis: Evaluation should cover what an agent is allowed to do, what information it can access and how a human can review or reverse an action. A successful demonstration on a tidy example is not the same as reliable performance on exceptions.
Microsoft sells AI-enabled business products, so its interpretation should be read with that commercial context in mind. The global survey also should not be treated as a South Africa-specific adoption rate.
The useful takeaway is a planning question: which work can be safely delegated, and which decisions still require accountable human judgement?