The paper introduces the concept of *designerly adaptation* for large language models (LLMs) to enhance user experiences. Designerly adaptation involves leveraging designers' unique perspectives and expertise to adapt LLMs to align with user needs and contexts. The authors conducted a formative study with 12 designers experienced in designing LLM-powered products to identify key characteristics of designerly adaptation, which include low technical barriers, the use of natural language prompts, and encouraging iterative tinkering. Based on these findings, they developed CANVIL, a Figma widget that supports structured authoring of system prompts, testing of adapted models, and integration of model outputs into interface designs. A task-based design study with 17 designers (6 groups) was conducted to investigate the implications of integrating designerly adaptation into design workflows. The study found that designers were able to iteratively tinker with different adaptation approaches, reason about interface affordances, and recognize the potential for collaborative adaptation. The work opens new avenues for collaborative processes and tools that emphasize designers' user-centered expertise in crafting and deploying LLM-powered user experiences.The paper introduces the concept of *designerly adaptation* for large language models (LLMs) to enhance user experiences. Designerly adaptation involves leveraging designers' unique perspectives and expertise to adapt LLMs to align with user needs and contexts. The authors conducted a formative study with 12 designers experienced in designing LLM-powered products to identify key characteristics of designerly adaptation, which include low technical barriers, the use of natural language prompts, and encouraging iterative tinkering. Based on these findings, they developed CANVIL, a Figma widget that supports structured authoring of system prompts, testing of adapted models, and integration of model outputs into interface designs. A task-based design study with 17 designers (6 groups) was conducted to investigate the implications of integrating designerly adaptation into design workflows. The study found that designers were able to iteratively tinker with different adaptation approaches, reason about interface affordances, and recognize the potential for collaborative adaptation. The work opens new avenues for collaborative processes and tools that emphasize designers' user-centered expertise in crafting and deploying LLM-powered user experiences.