Workflow, Not Models, Is the V-AI Moat
The real defensibility in Vertical AI isn't the LLM, but encoding undocumented industry workflows. An analysis argues that while 60-70% of the AI stack is commoditizing, the 30-40% focused on deep workflow integration—like Epic customizations in healthcare—is what commands $500K+ contracts. This is especially true in complex verticals like fintech and legaltech.
The legaltech AI startup Harvey is a prime example of the workflow moat in action. Instead of building a general-purpose AI, Harvey is trained on legal-specific data and designed to automate and augment high-value lawyer workflows like contract analysis, due diligence, and litigation prep. This deep integration into the legal process allows it to command high-value contracts from major law firms. Venture capitalists are increasingly prioritizing this vertical-specific approach. In Q4 2025, investors shifted focus to proven AI innovators with defensible business models, particularly in niche vertical solutions. In Q2 2025 alone, Vertical AI startups in the US and Canada saw 784 deals totaling $17.4 billion, with healthcare and financial services leading the way. Andreessen Horowitz, General Catalyst, and Tiger Global are among the most active investors in the space. For engineers looking to build, this means focusing on frameworks that excel at orchestrating complex, multi-step processes. Open-source tools like LangChain provide modular components for building LLM-powered applications, while alternatives like LlamaIndex are optimized for complex data retrieval and CrewAI is built for orchestrating multiple specialized AI agents. These frameworks are the technical foundation for encoding the complex business logic that creates a workflow moat. The NYC startup scene reflects this trend, with a surge in hiring for AI roles at companies embedding AI into specific industries. Companies like Hebbia (finance/legal), EliseAI (property management/healthcare), and various legal AI startups are actively hiring engineers. Even OpenAI is hiring AI Deployment Engineers in NYC specifically to help startups build novel applications on its platform. This focus on workflow provides a clear path for enterprise engineers wanting to build a side project. The key is to identify a high-value, repeatable process within a specific industry and build a tool that automates it. This could be a micro-SaaS for a niche in civil engineering or a tool to automate customer service workflows. The defensibility comes not from the base model, but from the deep understanding of a specific, often unsexy, business problem.
