Design · Boston, Massachusetts · FullTime · Senior level The role asks for a designer who shapes how an AI agent behaves, not just how it looks: its conversation, its handoffs to real work like quotes and invoices, its error recovery, and how it explains itself. A strong candidate has shipped LLM powered products with real users, writes prompts and system rules, reads traces to find where an agent went wrong, and turns findings into reusable patterns. Clear writing, systems thinking, and comfort in undefined, fast moving work matter here. How it fits the AI era The work sits directly on top of AI: the agent's tone, clarifying questions, confirmations, and recovery are design decisions written into prompts and rules. AI takes over much of the drafting and checking, while the human judgment stays in defining what good looks like, labeling real conversations, and deciding when the agent should act versus ask. The role also builds the evaluation habits that keep an agent trustworthy as models change. How it builds AI experience A candidate would gain hands on practice writing prompts and system instructions that ship, building rubrics and automated checks from labeled conversations, and prototyping against a live model rather than static screens. Because the company builds an agent for home service businesses, there is room to learn how non technical pros talk and where trust breaks, then turn that into patterns other teams can reuse. Worth showing Prompt and system writing Trace reading and diagnosis Rubrics and evals Prototyping in code This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role. About Topline ProAn AI platform that builds the website, answers the phone, writes quotes and sends invoices for small home service businesses.