The Harness, the Horse, or the Hay (queener.substack.com)

A recent analysis has highlighted a pivotal shift in the application landscape for AI, particularly in employee-facing software. The author argues that as organizations increasingly gravitate toward singular applications tailored for specific roles, around 75% of software companies in this space may become obsolete. This "one application" mindset implies a need for existing firms to redefine revenue models, as only one dominant player per function is likely to thrive. The analogy presented categorizes companies into three roles: the Horse (large language models), the Harness (application layers), and the Hay (support components). This distinction is critical for the AI/ML community as it illuminates the challenges ahead—particularly for startups competing in this landscape. While large model developers may tempt founders to pivot toward model development, structural barriers such as a lack of specialization and the risk of handing proprietary insights to model providers will likely deter such moves. Instead, focusing on creating specialized harnesses—deeply integrated applications that embody the ontology, interaction model, and data governance for specific user needs—remains essential. Companies that can deliver these comprehensive solutions will hold significant competitive advantages, emphasizing the importance of concentrated effort on defined roles rather than fragmented feature sets.