AI work redesign succeeds or fails at the task level, not the role level. The organizations getting real returns from AI are the ones who looked hardest at how work actually runs before they changed it. Jessica Neal, former CHRO at Netflix and now an investor and advisor to leading global enterprises, has seen this pattern play out twice: once during Netflix's pivot from DVD to streaming, and again now, as every major company faces a once-in-a-generation work transformation.
This conversation, from the first episode of Reejig's Work Blueprint series, covers what CHROs are actually worried about, why traditional change management is broken, and what it takes to lead the people function as a genuine P&L driver in the AI era.
The best-performing enterprises right now are doing one specific thing differently: assessing work through an AI-led versus human-led framework at the task level, not the job level.
Job-level analysis is too blunt. A role that looks automatable at the headline level may contain dozens of tasks, some revenue-driving, some critical for expertise development, and some genuinely ripe for automation. You cannot make those distinctions without task-level visibility.
As Neal observed, companies making progress have a senior coalition aligned on this framework: CHRO, CIO, Chief Product Officer, and CTO in the same room, with a shared agenda. Where that alignment breaks down, AI transformation fragments into a CIO-led tool deployment with low adoption, or a CHRO-led initiative with no engineering support.
One of the most underrated risks in AI work redesign is automating the wrong things. Every enterprise is deploying AI. Almost none can see the work they're deploying it into.
Siobhan Savage, Founder and CEO of Reejig, described the triangulation clearly: which tasks amplify revenue-driving work, which tasks cost money without creating value, and which tasks must stay human to protect the expertise infrastructure of the business. Removing developmental tasks from a role does not just save time. It hollows out the talent bench.
The examples are not theoretical. Government agencies that automated welder roles cannot now find welders. Retailers that removed bakery positions face critical expertise gaps. The pattern repeats. Work Architecture, not just job architecture, is the map that prevents it.
Enterprises are spending on AI and not finding the returns. In many cases the investment is real and the time savings are real. The problem is that no one told employees what to do with the time they got back.
If you automate invoice processing and free up two hours a day across two hundred people, those people fill that time with whatever they think is important. Intent is good. Information is not. The ROI disappears into unfocused activity.
Neal was unambiguous: employees don't know. Leaders are also not clear. The what and the why are not getting explained, so the how never reaches the workforce.
The fix is not a change management program. It is a communication structure at the role level: "Your workflow is changing. Here is what is different. Here is what we want you to do with the time you get back."
This is the core argument for Stealth Change Management. Continuous, embedded change delivered inside the systems your people already use. No kickoff meetings, no posters, no separate program. Just the work, updated. Work will upgrade like an iPhone upgrades. Getting organizations comfortable with that rhythm is the capability that makes every subsequent wave of AI redesign faster and cheaper.
The agent + human operating model introduces a management challenge most organizations have not yet processed: agents need to be managed, and that accountability lands on the humans working alongside them.
Most organizations have not updated their performance frameworks, their management development programs, or their hiring profiles to reflect this. The spans of accountability expand. Hierarchies become flatter. The most important capability shift is from task execution to judgment: the ability to direct agents well, evaluate their output critically, and own the consequences.
The argument Savage and Neal kept returning to: HR is not a compliance function managing the edges of an AI transformation someone else is leading. It is the function architecting how the organization works.
Neal framed it precisely. She never thought of herself as an HR person. She thought of herself as a chief effectiveness officer. Someone designing the organization around what makes people most effective, with a direct line to the P&L. Look at the tasks. Look at the structure. Look at the design. Get that right and the ROI impact is massive.
The roles emerging across large enterprises reflect this shift. A small workforce innovation team, what tech companies call forward-deployed engineers, drops into departments, maps AI opportunity, models agent cost versus people cost, and activates through the HR business partner network. Job architecture, once a compliance exercise, has become the architecture of how the company will operate in the AI era.
From Job Architecture to Work Architecture. That shift is the difference between a compliance catalog and the live infrastructure of an AI-powered enterprise.
|
CHRO Focus |
CIO Focus |
Shared Outcome |
|
Task-level work design and AI-led versus human-led assessment |
Agent inventory and AI stack governance |
A unified view of which workflows to redesign and in what order |
|
Employee communication at the role level: the what, why, and how of workflow changes |
Deployment governance and system integration |
AI investment that converts to measurable changes in how work runs |
|
Expertise pipeline protection: which tasks must stay human |
Cost-per-agent versus cost-per-person modeling |
AI adoption that builds organizational capability rather than eroding it |
Why are so many AI investments delivering zero return? Most AI ROI fails because tools are deployed on top of unchanged workflows. Without task-level visibility into how work actually runs, AI is applied to the wrong things. Consumption metrics, licences, prompts, logins, do not prove that work changed. Only actual changes to tasks and velocity do.
What does "AI-led versus human-led" actually mean in practice? It is a decision framework applied at the task level, not the job level. For every task in a workflow, the question is whether it should be executed by an AI agent, augmented by AI with a human in the loop, or kept fully human-led, based on complexity, judgment requirements, and expertise-pipeline value.
Why is task-level analysis more important than role-level analysis? Job titles describe what people are called. Tasks describe what they actually do. AI capability maps to tasks, not titles. A role that looks fully automatable at the headline level may contain dozens of tasks, some revenue-driving, some critical for expertise development, and some genuine automation candidates.
What is Stealth Change Management? Stealth Change Management is continuous, embedded work change delivered inside the systems employees already use, without kickoff meetings or separate programs. Traditional change management was designed for single, bounded transformations. AI-driven work change is continuous. Organizations must build the capacity to absorb ongoing change as standard operating rhythm.
How should enterprises think about automating too aggressively? Removing tasks that are also developmental pathways degrades the talent bench over time. Before automating any task, assess its role in building the expertise your organization needs five years from now. Someone has to monitor the agents.
What does the agent + human operating model mean for organizational design? Human roles become broader and less hierarchical. Fewer management layers are needed as agents handle more execution. The most important capability shift is from task execution to judgment: directing agents well, evaluating their output critically, and owning the consequences of what they produce.
The companies that get AI right are not the ones who moved fastest. They are the ones who looked hardest at the work before they changed it. Task-level visibility, a shared cross-functional framework, Stealth Change Management, and an HR function operating as chief effectiveness: these are the structural conditions for AI ROI that holds up under board scrutiny.
Book a demo to see how Reejig's Work Operating System maps your work at the task level and gives your team the AI Impact Analysis to start redesigning with confidence.