Startups and consultants race to trim enterprise AI spending with niche models

Enterprises are increasingly turning to a growing ecosystem of coaches, measurers and builders to rein in runaway AI costs. Oumi AI, founded in 2024, enables rapid creation of bespoke models, raising $10 million in seed funding. Consulting firm Adaptovate, with a global staff of over 100, now focuses on aligning AI strategy with talent and organizational design.

Measurement startup Larridin adds a data-layer that links token spend to employee output, securing $17 million from investors including Andreessen Horowitz. Infrastructure specialists Runware and Tensormesh supply cost-effective inference and caching solutions, backed by $20 million from hardware investors such as AMD and Nvidia’s venture arm. Emerging players like Conifer are also developing query-splitting technology aimed at cost-sensitive customers. Across the board, the advice is consistent: avoid using expensive frontier models for routine tasks, adopt lighter open-source alternatives, and treat AI spend as capital investment with a multi-year ROI horizon.