Where AI pays off in professional services firms
AI is now good enough to do real work inside a professional services firm. The hard part is not capability: it is choosing which work, and making sure somebody owns the change. In the engagements we run, the value clusters in three places.
Bid and proposal work
Firms live on documents assembled under time pressure from material that already exists. Drafting, restructuring and checking those documents against a house style is usually the fastest return, because a person still reviews the output before it goes anywhere.
Research over large document sets
Due diligence, discovery and regulatory review mean reading hundreds of pages to find the twenty that matter. AI does not replace the judgement in that work; it removes the reading around it.
Internal knowledge retrieval
Most firms already own the answer, somewhere, in a previous matter, proposal or policy. Making that knowledge findable is unglamorous and typically the highest-leverage change available.
Where it rarely pays off
Generic chat assistants handed round without a workflow, and pilots run by nobody in particular with no measure attached. Both generate enthusiasm and leave the way work gets done completely unchanged.
How we scope a first project
One workflow, one named owner, and one outcome you can measure inside a quarter. Six to eight weeks is usually enough to know whether a use case is worth scaling, and to stop cleanly if it is not.
Further notes
AI adoption · Sep 2026
Choosing the first AI use case
Why the first project should be a workflow that already has an owner, a review step and a measure attached, and how to tell a genuine candidate from a polished demo.
Knowledge management · Aug 2026
Making internal knowledge findable
Most firms already own the answer somewhere in a previous matter, proposal or policy. The work is in structure, permissions and how people search, not in buying another model.
Delivery · Jul 2026
Running a pilot that survives the quarter
A pilot needs a named owner, a defined output and a decision date. Without those three, momentum dies the moment the sponsor's attention moves elsewhere.