Every strategy engagement starts the same way: three to five days of desk research before a single piece of original thinking can happen. Industry landscape, competitive positioning, financial benchmarking, regulatory context, case studies. It's necessary. It's important. And it used to burn a full week of analyst time before the senior team could even frame the problem.
What changed
AI tools can now complete first-pass desk research in four to eight hours that previously took three days. The output isn't perfect — it requires review, triangulation against primary sources, and judgment about what's reliable — but the structural compression is real and it has materially changed what a team of five can accomplish in a two-week engagement.
The new workflow
Research in a modern consulting context now typically has three phases. First, AI-assisted synthesis: give the model the question, the scope, and the constraints. It produces a structured first-pass landscape. Second, gap analysis: review what the AI found, identify what's missing or uncertain, and prioritise the primary sources that need verification. Third, human synthesis: the insight layer — connecting the dots, identifying the counterintuitive finding, building the narrative — remains entirely human.
What this means for analyst development
The legitimate concern is what happens to analysts who used to develop judgment through the research grind. The research itself wasn't the point — the discipline of reading closely, finding the interesting signal in a mass of data, and synthesising under time pressure was. That skill needs to be developed deliberately now, not assumed to happen as a by-product of research tasks.
The quality question
Is AI research as good as analyst research? For factual synthesis and landscape mapping: often comparable, sometimes better. For primary source triangulation, nuanced interpretation, and identifying what's missing: worse, and probably always will be. The most effective teams are clear-eyed about which parts of the research question fall into each category.
The competitive implication is structural: a team that has integrated AI into its research workflow can outbid on engagements, deliver faster, or invest the recovered time into more original primary research. The question isn't whether to adopt — it's whether the speed advantage is passed to clients or retained as margin.