If you’re working on an AI agent right now that works perfectly for any given task, you’re making the wrong bet. AI model capabilities are only going to get better, context windows will get longer and have less rot, tool use will improve, compute efficiency will get better, and so on. The best thing to do is to be working on agents that can accomplish a smaller unit of work today, but necessarily require continued improvements to expand to broader tasks over time. This will mean you can get near term adoption and quick feedback now, but model gains will accrue to your agent so it steadily solve more complex work. A corollary to this also is if you’re in an enterprise and spending time evaluating agents that work perfectly today for all your use-cases, you may not be pushing them hard enough. We’re going to see major performance gains in the coming year, and you want to be in a position to get all that upside.