Building the future of agents with Claude

In this podcast episode, Alex leads a discussion with Brad and Katelyn about the Claude developer platform, formerly known as the Anthropic API. They discuss the reasons behind the name change, the platform's features, and its use for both external customers and internal products like Claude Code. The conversation covers the evolution of agents, the diminishing need for scaffolding as models improve, and the importance of auxiliary tools. They highlight the Claude Code SDK as a starting point for developers, even for non-coding applications, and emphasize the significance of identifying business value when targeting use cases for agents. The discussion also covers the importance of observability in long-running tasks and new features for managing context, such as removing unnecessary tool calls and agentic memory. Looking ahead, they express excitement about higher-order abstractions, self-improving outcomes, model launches, and giving Claude a computer to further unhobble the model's capabilities.