Anthropic's recent advancements, particularly with Claude Code and applications targeting specific industries, have sparked significant debate about the future of Vertical AI. This sector was initially seen as a promising opportunity for venture capital, focusing on tailored AI solutions within specific fields. However, as established foundation model companies quickly adopt vertical strategies, the landscape is shifting. Vertical AI firms need to assess their positions carefully. The ones that can thrive are those that effectively manage the last mile of AI, which includes delivering measurable outcomes while navigating complex regulations and real-world challenges. The article breaks down last mile defensibility into a spectrum, ranging from short tasks like content generation to long-term responsibilities such as tax compliance and healthcare oversight. Shorter last miles face more competition from foundation models, while longer last miles provide stronger barriers against these competitors. Companies must think critically about their roles: who holds the liability, who manages relationships with regulators, and who ultimately owns the customer outcome? A firm that answers these questions affirmatively for itself is likely to enjoy a more secure position in the market. Owning liability is a key theme. Companies that take responsibility for outcomes can command higher trust and pricing power. For instance, a basic tax software tool is less valuable than a comprehensive service that handles filings and audits. The article argues that long last mile coordination is not just about filling gaps with human intervention; it’s also about leveraging unique insights and superior technology to create a more robust offering. Examples like Outpost, which manages global tax compliance, illustrate how companies can integrate complex systems to enhance their value proposition and fend off competition effectively.