What the Research Tells Us
AI adoption follows a J-curve pattern. In an initial phase, firms experiment and workers learn, but the impact on productivity is modest because the existing organisation of work was not designed around AI. Then comes deeper change: firms re-engineer processes and production structures to make the most of the new technology. This is when the large productivity gains—and the large disruptions—arrive.
At the firm level, this plays out through creative destruction. Successful adopters gain market share; slow movers lose it. New entrants built around AI from the start can rapidly displace incumbents. The aggregate patterns of displacement and reinstatement emerge from winners and losers at the firm level, not from uniform adjustment across industries.
While the dynamics at the industry or economy level are very hard to predict, a lot more can be said about how AI can be deployed at the task level today. This is the starting point for evaluating how AI impacts your organisation.
For a deeper examination of the research on which our approach is based, read about our broader framework here.
Our Approach
We start by stepping below the level of jobs to look at the underlying tasks a firm needs to undertake. Tasks have been bundled into jobs historically to trade off the benefits of specialisation against the costs of coordinating across separate roles (for a fuller treatment, read the longer blog). AI not only automates and augments tasks, but also changes the economics of this trade-off in a way that means we need to re-think how we structure work in our organisations.
We use a structured interview process—delivered via an AI assistant—that maps which tasks employees perform, how those tasks depend on one another, who does them, and which systems they interact with. This builds a detailed network graph of the tasks in your organisation: not an org chart, but a picture of how work actually gets done.
This network graph reveals patterns that org charts miss. Some tasks sit at coordination hubs, with many dependencies flowing through them. Others are peripheral. Some require tight interdependencies with specific other tasks; others stand alone.
We then apply the task-based research to your specific structure, evaluating opportunities across four channels:
Task automation: AI performs a task entirely, removing it from the human
workload.
Task augmentation: AI assists with a task, raising quality or speed while the
human remains in the loop.
Coordination compression: AI reduces the cost of splitting interdependent
tasks across people, enabling tighter specialisation.
Skill-requirement reduction: AI lowers the skill threshold for certain tasks,
changing which roles make economic sense.
While automation, augmentation or coordination compression may save time at the task level, turning that into productivity gains requires re-organising tasks at the job level to free up time in a way that is most valuable to the firm. This in turn requires understanding the dependencies between tasks, which is where the task graph comes into its own.
Understanding risk
Not all opportunities are equal. Automating a peripheral task might be a safe quick win with modest productivity gains. Automating a coordination hub could be transformative—but runs the risk of serious disruption if it goes wrong. Understanding the structural position of each task is essential to sequencing adoption well. See the detailed methodology for a fuller explanation and a worked example.
This work draws on the latest academic research on task-based production, combined with proprietary data on your specific organisation. We keep tabs on the evolving evidence and bring it to bear on the questions you face. Track the research here.
What You Get
An assessment of where you could use AI in your organisation and guidance on how to get started, comprising:
- a task graph mapping of how your work is currently structured;
- opportunity identification across automation, augmentation, coordination and skill compression channels;
- a risk assessment and adoption sequencing recommendations;
- an implementation roadmap tailored to your priorities.