What Harvey does well.
Harvey is the leading purpose-built legal AI assistant. Trained on legal data, deeply tuned for AmLaw-segment workflows: deal-team research, due diligence document review, complex contract analysis, regulatory research. At firms with 200+ attorneys and an Innovation Partner who can drive adoption, Harvey ships meaningful value. The product is real, the team is excellent, and the AmLaw market voted with their wallets.
For 20-150 attorney mid-market firms, Harvey is sometimes the right answer and sometimes the wrong one. The factors that determine which are different from the AmLaw factors.
Where Harvey is the right answer.
Three patterns where we tell mid-market firms that Harvey is worth evaluating:
The firm whose practice is heavily document-review-bound. M&A due diligence, large litigation discovery, complex contract review. Harvey is purpose-built for this and the off-the-shelf product covers most of what the firm needs.
The firm whose attorneys would actually use a separate AI interface. Adoption matters. Some mid-market firms have culturally adopted Lexis+ AI or Westlaw Precision and adding Harvey alongside fits the muscle memory. Others have not, and Harvey's interface becomes a barrier.
The firm with formal Innovation budget. Harvey is priced for firms with structured tech budgets. Mid-market firms without that often balk at the per-seat economics, especially when most attorneys won't use it daily.
Where Harvey is the wrong answer.
Three patterns where we tell mid-market prospects that Harvey is wrong-segment for them:
The firm whose constraint is partner-hour leakage in time capture, intake, and matter summarization, not document review. Harvey doesn't address billable-hour reconstruction, doesn't write to iManage Time, doesn't do AI intake triage. The 47-attorney firm with $1.4M in annual unbilled-time leakage cannot solve that with Harvey. The iManage AI Integration Playbook describes what does.
The firm running iManage or NetDocuments who needs the AI inside that DMS, not in a separate tool. Harvey is a separate interface. The leverage at most mid-market firms is in the AI running invisibly inside the system attorneys are already in. Custom AI on top of iManage or NetDocuments writes into the workspaces attorneys already use.
The firm whose AI use case is firm-specific knowledge retrieval over decades of memos and matters. Harvey's RAG is tuned to public legal corpora. Your firm's institutional knowledge is private and lives in your DMS. Custom retrieval over the firm's archive, with permissions intact, ranks among the highest-leverage commissions we run.
The honest side-by-side.
Harvey's strengths: mature product, deep legal training, strong support, AmLaw-validated, fast off-the-shelf deployment for document-review use cases, well-funded roadmap.
Custom-commission strengths: built into the firm's actual DMS (iManage, NetDocuments), reads the firm's actual archive with permissions intact, addresses the specific leakage workflows Harvey doesn't (time capture, intake, matter summarization), owned by the firm at handoff.
Harvey's weaknesses, for the mid-market: separate interface attorneys must adopt; per-seat pricing strains mid-market budgets; off-the-shelf training doesn't see the firm's institutional knowledge; doesn't address the leakage workflows where mid-market firms have the most exposure.
Custom-commission weaknesses: bigger one-time spend; longer initial scope; only worth it for firms with workflow specifics; off-the-shelf legal-research alternatives (Lexis+ AI, Westlaw Precision) cover most generic legal-research needs better.
What we recommend.
If your firm's biggest leverage point is document review at scale, Harvey is worth a real evaluation. If it's the time capture, intake, billable-hour reconstruction, or knowledge retrieval inside your iManage or NetDocuments archive, the answer is custom AI on top of your DMS, not Harvey. Run the Billable-Hour Recovery Diagnostic to find out which.
About 50% of the law-firm diagnosis calls we run end with us recommending an off-the-shelf tool (Harvey, Spellbook, Lexis+ AI). The other 50% end with a custom-commission scope. The honest framing: the right answer depends on the workflow, the stack, and the firm's specific leakage profile.