Vertical AI leaders Harvey (legal) and Sierra (CX) are succeeding by focusing on specific, high-value enterprise workflows and building deep domain expertise.
Product development in AI requires extremely agile, continuous planning cycles (monthly to quarterly) to adapt to the rapid evolution of foundation models.
Deep customer collaboration is paramount, utilizing strategies like forward-deployed teams, sharing roadmaps directly with clients, and co-developing solutions with 'frontier customers'.
User experience (UX) and domain-specific adaptations (e.g., local dialects, industry-specific workflows) are critical, underrated differentiators that compensate for model weaknesses and drive user adoption.
9 quotes
Concerns Raised
The rapid pace of foundation model development can render application-layer features obsolete overnight.
Maintaining team morale and focus amidst constant, rapid-cycle planning ('thrash') is a significant operational challenge.
Building trust with enterprise customers around model reliability and data security for on-premise data remains a hurdle.
Opportunities Identified
Unlocking immense value from proprietary, on-premise enterprise data that foundation models cannot access.
Expanding from individual productivity tools to multi-player, organizational platforms that create network effects.
Addressing customer demand for 'long horizon tasks' like drafting entire complex documents or managing multi-turn customer relationships.
Using superior UX and deep workflow integration as a durable competitive advantage against both incumbents and new entrants.