LESSONS OF FIN FOR AI ASSISTANTS
In 2015, Fin's team tried to build an AI assistant service and found problems that even today's better AI can't fix. While they built good systems for remembering user preferences and checking service quality, they found a basic issue: when you stand between users and real-world services, users blame you for every problem. Even with better AI today, assistant services still face the same basic challenges that made Fin change its business model.