To justify the high cost of AI apps... they must prove 'work outcomes' rather than performance

a16z’s November 2 “How AI Applications Earn Their Premium” newsletter argues that AI app pricing must be justified by measurable work outcomes rather than model performance alone, citing legal teams that saved four to six hours per week using legal AI service Harvey. The firm said vendors need to prove they outperform customers’ alternatives and price against that value, because falling inference costs improve margins but easier model and tooling access also makes in-house builds more feasible. The piece points to Harvey’s deployment at law firm Latham & Watkins as an example of outcome-based justification. The takeaway for AI software vendors is that defensible premiums will depend on execution and information leverage, not raw answer generation. What to watch next is whether AI app makers tie pricing to documented productivity gains as general-purpose models close the capability gap.