Analytical work is one of the clearest examples of why “AI can do the job” is the wrong framing. The job is not one thing.
An analyst or consultant may need to define the question, collect information, clean data, understand context, test hypotheses, calculate, compare alternatives, write a narrative, build a presentation, revise it after feedback and defend the conclusion. Different parts of that chain have very different AI potential.
Compress the work around judgment — not the judgment itself.
Research and data work
AI can accelerate discovery, summarize large volumes of material, extract structured facts and identify themes. But the analyst still needs to judge source quality, relevance, comparability and whether the evidence supports the conclusion.
The same is true for data work. AI can help write transformations, inspect schemas, generate code, explain errors, propose validation checks and produce first-pass analyses. This can be extremely valuable even for professionals who are not programmers.
The point is not to turn every analyst into a software engineer. It is to reduce the friction between a question and a reproducible calculation.
A first draft is cheaper than a blank page
If the analytical structure is already clear, AI can generate a first narrative, convert notes into an executive summary, restructure material for different audiences and produce alternative formulations. But the analyst should remain responsible for the argument.
AI is particularly useful for iteration. Instead of spending time on mechanical rewrites, you can quickly stress-test the work with questions that might otherwise appear much later in review.
These are high-frequency microtasks that can materially shorten the path from working material to decision-ready output.
Because AI can produce fluent explanations quickly, it can make weak analysis look finished. Calculations should be reproducible, sources traceable, important claims checked, and uncertainty kept visible.
What to remove from an analyst’s day
- repeated searches for the same context
- manual restructuring and repetitive formatting
- copying between tools
- long delays between an idea and a first test
- administrative effort around revisions
The right objective is not to automate the analyst out of the process. It is to remove the parts of analytical work that do not require the analyst’s scarce judgment.
The human then spends a larger share of the day doing what clients and organizations actually value: asking the right question and deciding what the evidence means.