Management: AI trends to focus on - Redesigning work, not just distributing AI tools

This week the conversation moved from “give people AI tools” to “redesign the work itself.” The evidence shows adoption is still patchy, ethics codes are silent on deployed AI, and managers are being asked to define decision rights—not just distribute licenses.

What changed this week

Three signals converged. First, the pattern of workflow redesign became explicit. Multiple sources pointed to the same practical sequence: define the process, assign who decides what, monitor outcomes, automate repeatable work, and keep humans on exceptions and high-consequence choices. This is not a technology problem. It is an operating-model problem.

Second, the gap between organisational intent and frontline reality widened. Workday’s research on frontline workers found a widening expectations gap on schedule flexibility, pay clarity and access to digital tools. A separate academic study confirmed that workplace AI adoption remains informal and uneven. Managers are pushing forward while many employees are still uncertain about what the tools do and who is accountable for the output.

Third, governance moved closer to the workflow. SiliconAngle reported that AI governance is shifting from policy documents to in-the-moment controls. TheCube’s analysis described governance that sits inside the tools people use, not in a separate compliance function. At the same time, the Work Futures report found that most workplace ethics codes remain silent on deployed AI. That silence is becoming a liability.

Two industry-specific moves reinforced the theme. Contact centres are shifting from generative answers to agentic resolution—systems that act, not just respond—while physical AI is targeting fragmented hospital logistics. Both cases require managers to decide what the machine can complete and what must stay with a person.

What it means for you

You are being asked to redesign work, not just add AI. That means sitting down with a specific process—scheduling, approvals, customer handoffs—and drawing a line between what runs automatically and what requires human judgment. If you cannot point to that line, your team cannot either.

Adoption will stay uneven until you make decision rights visible. People need to know: “The AI handles this step. I handle this one. Here is where I escalate.” Train on real workflows, not generic AI literacy. The Workday and academic findings both suggest that abstract training does not close the expectations gap. Concrete, role-specific practice does.

Governance is now your job, not just legal’s. When ethics codes are silent and governance moves into the workflow, the manager becomes the first line of accountability. If an automated decision affects a customer or an employee, you need to know who reviews it and how. The banking sector’s call for human accountability applies to every industry running agentic systems.

The workforce replacement conversation is getting louder. A manager survey from Aiven Consulting Group showed rising support for AI-led workforce reduction. Even if that is not your intent, your team is reading the same headlines. Be direct about what is changing and what is not. Silence will be read as confirmation.

What to focus on next week

• Pick one end-to-end process your team owns. Map the steps and mark each as “automate,” “human required,” or “human reviews.” Share it with the team and ask for corrections.
• Run a 30-minute session on a real workflow that now includes AI. Do not present slides. Walk through the actual tool, show the handoff points, and let people practice the exception path.
• Check your organisation’s code of conduct or ethics policy. Search for “AI,” “automated decision,” or “algorithm.” If you find nothing, flag it to your leadership and ask what guidance applies to your team’s tools.
• For any agentic or automated system your team uses, identify the human who reviews its decisions. If no one is named, name someone this week—even if temporarily.
• If workforce change is being discussed above you, prepare three facts about your team’s current output, the work that only humans can do, and the cost of getting automation wrong in your context. Have them ready before you are asked.

These briefings draw on the full set of stories we track each week. For every article summarised above and more, see all Management AI news.