AI doesn't map to job descriptions. It maps to decisions, outputs and tasks. Most organisations only hold job-level data: roles, responsibilities and reporting lines. Without task-level data on how work actually breaks down, you can't design human-machine workflows. You bolt AI onto the edges and hope.
The organisations getting value from AI map the flow of work first.
Why job descriptions don't tell AI where to help
A job description says "analyse data, prepare reports, coordinate stakeholders". That worked when people did everything. AI doesn't care about job titles. Within "analyse data" there might be twenty different activities: data cleaning, pattern recognition, interpretation, recommendation. Some are routine outputs AI handles well. Some are judgement calls that need human expertise. Some are decisions that could be partly automated with human oversight.
You can't just "add AI" to a job. You have to know what the job actually contains, and most organisations don't. They have job descriptions written years ago, org charts showing reporting lines, role levels and salary bands. What they lack is data on how work processes break down.
What organisations can't answer about their own work
Where does time go? Which outputs are routine? Which decisions need expertise? Which tasks involve judgement that's hard to codify? Most organisations can't answer these questions with data. They would have to ask people, and even then people often can't explain how their work actually flows.
So when AI arrives, there's nothing to design against. Someone says "let's use AI for research". Which parts of research? Literature review, data gathering, synthesis and recommending conclusions are different activities. Some are routine outputs, some are judgement-heavy decisions, and AI handles them differently. Without understanding the process, you're guessing.
Why AI ends up at the margins
The result is AI at the margins: a bit of efficiency here, some time saved there. The underlying work processes stay the same, with AI bolted on. That isn't transformation. It's automation at the edges.
The real opportunity is redesigning work around human-machine collaboration. People focus on judgement, relationships and complex decisions. AI handles routine outputs, data processing and pattern recognition. You can't design that split without knowing how work breaks down into decisions, outputs and tasks. The constraint isn't AI capability. It's data about the work itself.
What the organisations getting value did first
Organisations getting real value from AI did something unglamorous first: they mapped their work processes. Not job descriptions, and not process diagrams of the ideal state, but actual decisions, outputs and tasks. Where time goes, and what each step depends on. Then they could make explicit choices. This decision stays human. That output gets AI support. This task is fully automated, that one partly. Designed, not guessed.
Most organisations skip this step. They buy AI tools and hope for transformation, and get incremental efficiency instead. Work processes designed decades ago for people doing everything will hold back any AI added on top.
The question isn't whether your AI is good enough. It's whether you understand your work well enough to know where AI fits. That understanding is data in its own right: process-level knowledge of how work actually happens, the work context AI depends on. Most companies don't have it. Until they do, they're adding AI to work designed in 2006 and wondering why nothing changes.
