Founders face a strategic trade-off depending on the market cycle. In a hot market, capital is abundant but competition for user attention is fierce. In a quiet market, capital is scarce, but it's easier for a quality product to stand out and get noticed.
For tools like Harvey AI, the primary technical challenge is connecting all necessary context for a lawyer's task—emails, private documents, case law—before even considering model customization. The data plumbing is paramount and precedes personalization.
Harvey is seeing a powerful network effect where enterprise clients demand their outside law firms purchase Harvey to collaborate more effectively. This creates a highly efficient, low-cost customer acquisition channel driven by the end customer.
Traditional, static benchmarks for AI models go stale almost immediately. The superior approach is creating dynamic benchmarks that update constantly based on real-world usage and user preferences, which can then be turned into products themselves, like an auto-routing API.
OpenRouter's CEO views new model releases as marketing events. Users form personal attachments to specific models and actively seek out apps that support them. This creates recurring engagement opportunities for developers who quickly integrate the latest models.
True defensibility comes from successfully navigating successive challenges that weed out competitors. Many have an idea, fewer can build it, even fewer can maintain shipping cadence and distribution, and only a handful can raise capital at scale, leaving a 2-3 horse race.
The value of an AI router like OpenRouter is abstracting away the non-technical friction of adopting new models: new vendor setup, billing relationships, and data policy reviews. This deletes organizational "brain damage" and lets engineers test new models instantly.
A key job for junior lawyers is understanding non-legal context for a case, like a pharmaceutical supply chain. AI excels here by rapidly synthesizing massive amounts of diverse, industry-specific information alongside legal precedent, which is a core part of the value.
AI models reason well on Supreme Court cases by interpolating the vast public analysis written about them. For more obscure cases lacking this corpus of secondary commentary, the models' reasoning ability falls off dramatically, even if the primary case data is available.
Harvey AI's co-founder predicts AI will allow law firms to break the traditional billable-hour model. This shift will enable them to operate at a much larger scale with software-like margins, fundamentally changing the industry's structure and creating massive winners.
Instead of internal testing alone, AI labs are releasing models under pseudonyms on platforms like OpenRouter. This allows them to gather benchmarks and feedback from a diverse, global power-user community before a public announcement, as was done with Grok 4 and GPT-4.1.