Most workforce plans start with the wrong question: How many people do we need? That’s a headcount question, and no amount of sophistication can make it answer what matters now. The question that matters is: What work creates value, who should do it, and what should AI handle instead? This article covers both halves of the answer: planning the workforce, and redesigning the jobs inside it.
The Assumption That Broke First
Traditional workforce planning assumes jobs are fairly fixed units of work, and planning means figuring out how many units you need. AI broke that assumption. When AI can handle a large share of the tasks in a knowledge-heavy role, the job is no longer the right unit to plan around. The task is. Which parts should a person do? Which should AI do? Which should a person and AI do together?
Almost no role will disappear entirely. Almost every role will look different. So the useful question isn’t “which jobs will AI eliminate?” It’s “which tasks in every job can we redesign with AI, in a way that creates value and keeps people engaged in the work that matters most?”
“The goal isn’t to use AI to do the same work with fewer people. It’s to do better work: more value for customers and more meaning for employees.”
Five Questions Your Workforce Plan Should Answer
Here’s the framework I’ve found most useful with mid-market leadership teams. Five questions, in order.
What work creates value, and what doesn’t?
Start with the work, not the org chart. Where are the bottlenecks? Which roles generate margin and which absorb it? What exists only because of legacy process? Do this before any hiring or restructuring decision.
What capabilities does your strategy actually require?
Not the ones you have; the ones you’ll need in two to five years, given your market and what AI will be able to do. Which skills are rising, which are fading, and which can you build internally?
Build, buy, borrow, or automate?
For each gap: hire, develop, partner, contract, or automate? The right answer varies by role, timeline, and cost. Defaulting to “hire” is what inflates headcount without improving results.
What does this do to your economics?
Labor cost scenarios, revenue per employee, margin under different setups. If workforce planning isn’t connected to the financial model, it’s wishful thinking.
Can your organization actually absorb the change?
The question almost everyone underweights, and the one that sinks the most transformations. Leadership maturity, decision speed, and willingness to experiment aren’t soft factors; they’re execution risk. Our Change Capacity Index measures them.
Redesigning Jobs Task by Task
Once the plan points to the right work, the next step is redesigning the jobs themselves. Ravin Jesuthasan’s task deconstruction approach gives a practical structure in four steps:
- List the tasks. Not the job description, which is almost always an abstraction, but the work people actually do day to day.
- Assess what AI can do. For each task: can AI assist, fully automate it, or help a person produce a better result?
- Identify what stays human. Tasks that need judgment, empathy, ethics, or relationship depth should get more of your people’s time, not less.
- Redesign the role. Combine the human-essential tasks with new AI-assisted capabilities into roles that are more valuable, more engaging, and more productive.
After that, work usually falls into three groups:
- AI-automated: data processing, routine documents, scheduling, spotting patterns in large data sets. The person’s role shifts to oversight and quality.
- AI-augmented: research, analysis, drafting, synthesis. People still do the work, much faster and more consistently.
- Human-essential: strategic decisions, complex negotiations, hard conversations, original creative work. AI frees up time for these.
The best companies deliberately shift their people toward the third group while AI handles the first two more efficiently than was ever possible.
Task deconstruction is a value-creation exercise, not a cost-cutting one. If you’re using it mainly to reduce headcount, you’re misusing one of the most powerful design tools you have. Our Job Redesign Studio walks your team through it.
Tools: start with your highest-volume tasks
Three kinds of tools are already proving their value in mid-market companies: workflow automation platforms that connect systems and automate multi-step processes without custom code; AI assistants built into the tools your people already use; and specialized industry tools for sales, service, finance, HR, legal, and supply chain. Start with the most time-consuming, highest-volume tasks, and pick tools for those. Don’t let the number of options stall the decision.
What It Looks Like in Practice
I worked with the leadership team of a professional services firm: about 200 people, growing fast, and feeling like they were running harder just to stay in place. Their workforce plan was essentially “hire more of the people who got us here.”
Over six weeks we mapped the actual tasks, not the org chart, and asked the AI question for each: human-led, AI-led, or both together? Then we connected their three-year growth strategy to the specific skills they’d need. The result wasn’t a headcount target. It was a redesign of how three core service lines got delivered. Two roles that had been headcount bottlenecks turned out to be automation opportunities. Two capabilities they’d been trying to hire externally turned out to be buildable internally, faster.
Margin improved. Recruiting pressure dropped. The team stopped feeling like the business was outrunning them.
Workforce planning software (skills taxonomies, AI forecasting, internal talent marketplaces) is advancing quickly and can help. But software alone almost always falls short. Workforce planning is a management, governance, and capital-allocation discipline. The technology surfaces the data; leadership has to decide what to do with it, and build the system that acts on it consistently.
Five Principles for Intelligent Job Design
- Design for value, not just efficiency. More value for customers, the business, and employees, not just the same work for less.
- Involve the people doing the work. They know best which tasks can change. Include them in the design, not just the rollout.
- Redesign upward. Free people for work that is more strategic, creative, and meaningful.
- Move fast enough to learn. A good-enough plan plus quick iteration beats a perfect plan. Design sprints and Lighthouse Projects help.
- Treat change capacity as a design limit. Sequence the redesign around what your people can actually absorb, not just what the technology allows.
The companies that sustain performance through AI disruption aren’t the ones with the best plan. They’re the ones with the most capacity to keep changing. If your AI investment isn’t reaching the bottom line, or your workforce feels out of step with your strategy, a 30-minute conversation is usually enough to see whether there’s something useful here.