AI Orchestration and the Middle Management Transition

How can we best deploy the experience and skills of managers whose roles are at risk from AI disintermediation to help create a new, smarter work coordination system?

Anthropic CEO Dario Amodei took a lot of flak when he broke ranks recently to warn people that a significant proportion of entry-level white collar jobs could be wiped out by AI. But a more precise way to think about this shift is to consider jobs as bundles of tasks, and to think in terms of how AI will unbundle and rebundle those tasks into human activities, human-guided AI activities and tasks that can be fully automated.

We have argued for a long time that excess management layers are a very expensive drag on organisational performance, and often a barrier to talented contributors achieving their full potential. Instead of using technology to reinforce the old structures of process management and top-down control, we should be using it to elevate and connect the work of contributors to create more agile, autonomous organisations. And AI looks like it could be the best catalyst we have ever had to make this a reality.

So whilst it is true many entry-level tasks will be automated, organisations that can benefit from fresh talent and ideas should be able to re-shape or re-bundle job roles to account for task automation.

But for those managers who mostly perform coordination and oversight roles, there is a good chance that much of this work can be done better by AI, which will quietly and efficiently absorb the visible, repeatable parts of the role, such as reporting cycles, task allocation, weekly updates, and much of the spreadsheet logic that keeps large organisations running.

This should be seen as a moment of opportunity — both for the organisation to shed vastly inflated bureaucratic costs; for individual contributors, who will be more free to shape their roles; and also for the managers themselves, who can put their experience and knowledge to better use.

Management layers have acted as the glue between strategy and delivery. But much of that glue has been operational filler - compiling notes, translating KPIs into updates, formatting status decks, and other ‘busy work’. With agentic AI now handling co-ordination and monitoring, managers have the chance to move into more valuable territory — away from micromanagement and toward deeper judgement, orchestration, coaching and culture-shaping.

As the AI-inflected art project Uncertain Eric puts it:

AI eats legibility…anyone whose job involves moving language of logic around a computer. That’s the target vector. Not because they’re expendable, but because they are legible.

Managers are not being replaced. They are being refocused. The ones who engage with this shift will become designers and architects of how work happens, not just enforcers of what gets done.

What’s Actually Changing in the Work of Management

The daily reality of management is already shifting. Tasks that once demanded human oversight will increasingly be handled by agentic systems operating in the background, co-ordinating, tracking and even surfacing insights without needing constant supervision.

Here are some examples of the shift:

• Task co-ordination: AI can now assign work based on availability, priorities, and role context. This reduces the need for manual delegation.

• Progress tracking and reporting: Instead of managers compiling updates, agents can extract key milestones from tools and channels automatically.

• Risk identification & escalation: Predictive analytics and real-time monitoring allow AI to flag problems before they become visible to a human.

• Information synthesis: Summarising meetings, creating status reports and analysing sentiment from team interactions can now be handled by prompts and agents.

These functions are not disappearing, but they are no longer solely the domain of the manager. What remains is the work AI cannot do well: making sense of context and ambiguity, managing team dynamics, resolving unstructured problems, and the ‘world building’ needed to develop the best possible workplace culture.

Re-wiring our organisations for the future

The operational load is being lifted. What remains is the leadership load, which demands new capabilities, mindsets and habits that are better suited to managing complexity, enabling human potential, and working in partnership with intelligent systems.

Signals from the Edge: AI as the Operating Core

Venture capital is backing a new generation of businesses that put agentic AI at the centre. One recent example is early OpenAI investor Elad Gil, who recently announced his latest focus: AI-powered rollups. These are companies that acquire multiple small or mid-sized businesses, then run them more efficiently using AI agents, not just for support functions, but for core operations and management.

In these models, agents handle pricing, marketing performance tracking, reporting, and (at least partially) customer service. Humans remain in the loop, but they are no longer the primary operating layer. They are advisors, reviewers and strategic guides. The AI stack runs the basics.

In this kind of vision, prompt literacy, orchestration skills, and systems-level thinking are no longer niche capabilities - they are becoming core leadership competencies.

From Workflow Manager to Prompt Strategist

Most middle managers using GenAI today are doing so informally. Prompts live in personal chat histories or are shared ad hoc in meetings and messaging channels.

Instead of viewing prompts as clever one-offs, we can link them directly to the Jobs To Be Done (JTBD) that managers face every week. For example:

• Preparing a team update? Use a reusable prompt to summarise notes, progress, and blockers across tools

• Writing a performance review? Feed in structured input and tone guidance to generate a draft

• Analysing customer feedback or internal sentiment? Prompt the system to extract patterns from large volumes of text

• Planning a retrospective or kick off? Generate agenda options based on context and team priorities

This shift turns GenAI into a proper management assistant. Prompting becomes a design activity - modular, repeatable, and aligned to the rhythms of the work - that also encourages us to ask the right questions and express our goals with clarity.

Prompting becomes the new briefing. And learning how to prompt strategically becomes a new form of leadership leverage.

Read on to find out how managers are building prompt memory, designing automations and leading smarter systems, one task at a time.