Many ai agents are reportedly failing in production, with some outlets suggesting this is due to being built backwards or having a context problem. The reasons for these failures are not entirely clear, but possible issues include difficulties at the deployment stage, a lack of trust, and high costs. Some companies are introducing new methods and tools to improve the effectiveness of ai agents, but the overall success rate of these agents remains uncertain.