Work, reworked: What it takes to win in the age of Agentic AI

It’s often said that if ancient cities were built today, they’d look entirely different—new street grids, transit systems, and infrastructure. Yet for practical reasons, cities work around outdated foundations. In the context of agentic AI, most organisations resemble those old cities—constrained by busywork, inefficient processes, legacy structures, and cultural inertia. Introducing AI agents without rethinking these foundations risks replicating inefficiencies at scale. Incremental adoption can end up adding an agentic layer to an outdated way of working.

Making the leap: 5 key shifts to enable
the Human-Agentic Workforce

Can humans and AI agents really work together as a team? Absolutely—in fact, soon it will be an imperative for competitive advantage, given how quickly agentic AI capabilities are evolving. But it won’t happen until organisations reassess, rethink, and reengineer traditional models for how work gets done.

If that sounds like a big deal, it is—but it’s also entirely within reach. It will require confronting outdated paradigms of control, capacity, and incremental change. Laying the groundwork for human-agentic change starts with these five shifts:

• Technology‑led transformation → Human‑centric transformation
  Treat agentic AI as workforce and business transformation, not an IT project. Design for human impact and let technology enable amplified human potential.
• Task‑based roles → Outcome‑based design
  Replace fragmented tasks with dynamic, skills‑based roles organised around outcomes and human–agent collaboration, freeing people for judgement, empathy and creativity.
• Digitised processes → AI‑native business and operating models
  Move beyond adding AI to legacy systems: build AI‑native, fluid operating models that scale impact rather than amplify inefficiency.
• Headcount‑fuelled growth → Scaled performance through redesign
  Growth should come from reimagining work and orchestrating human–agent workflows, not simply adding people.
• Cultural inertia → Organisational readiness
  Address fear, siloed governance and low trust. Readiness, trust, governance and learning cultures, is a strategic differentiator for safe, rapid scaling.

These are not simple mindset shifts. Instead, they are systems-level transformations in how enterprises think, lead, organise, and perform—and none can be solved in isolation.
 

Beyond cross-functional collaboration:
Systemic codependency 

No single domain can act in isolation in an agentic AI world—not technology, operations, talent, customer, finance, risk, or legal. This means that cross-functional collaboration—typically collaborating across silos—isn’t sufficient for enabling successful agentic AI contributions at scale. Systemic codependency, in which progress in one area is only possible through alignment and shared ownership with others, is needed. Each part of the enterprise is a mutually dependent driver of AI-enabled progress.

How does this work in practice? These three actions provide the most direct path to systemic codependency:

• Reimagine leadership and governance: CIOs, CHROs, CROs and business leaders must operate as an aligned leadership system, shifting focus from managing people to managing outcomes enabled by human‑agent teams.
• Redesign work from the ground up: deconstruct legacy roles and workflows and reassemble them into outcome‑based, role‑and‑skill ecosystems for fluid human–agent teaming.
• Build organisational readiness as a strategic differentiator: Organisations risk scaling dysfunction if they simply add agentic AI; AI is evolving from rules‑based automation to generative AI and now to agentic, human‑on‑the‑loop systems. It functions at two levels:

Individual: boosts productivity and delegates multi‑step tasks to agents.
Enterprise: orchestrates end‑to‑end workflows but requires redesigned roles, decision rights and readiness.

Organisational readiness must be developed alongside innovation: workers must feel equipped, empowered and valued with AI as a teammate, while enterprise infrastructure, policies, leadership and culture evolve to absorb and scale agentic impact. Close the gap with a dual‑speed approach, rapid experiments where readiness exists (fast lane) alongside enterprise‑grade guardrails for trust, governance and alignment (stable lane), and turn these concurrent efforts into a coordinated plan for decisive action.
 

The Human‑Agentic Workforce in action

Lemonade built an AI‑native insurer: agents like “Maya” (onboarding) and “Jim” (claims) are core to operations. Agents execute high‑volume tasks autonomously while humans manage exceptions, quality and strategy. By organising work around outcomes and skills, not by bolting AI onto legacy roles, Lemonade scaled performance without proportional headcount growth, embedding governance, transparency and continuous learning from the start.
 

No more experiments. It's time to scale.

Move from fragmented pilots to enterprise‑wide transformation. Start where you stand by clarifying your AI ambition, acceptable autonomy and readiness. Use a dual‑speed model to balance rapid innovation with stable guardrails. Pivot from vision to impact by defining baseline requirements, linking agents to measurable business outcomes, ensuring data and workflow readiness, setting human–agent autonomy thresholds, securing high‑impact quick wins and scaling proven approaches across the enterprise.

Not next quarter or next year - now

AI is here now. The choices leaders make today will determine whether AI becomes complexity or a catalyst for a more human, intelligent and productive system of work. Move from pilot paralysis to enterprise reinvention and architect a human‑agentic future where humans and agents together deliver outcomes neither could achieve alone.