Defining the Agent's Environment › Environments Evolve

Here's something important to remember: your agent's environment isn't static. It grows and changes as you add new capabilities.

When we first built our assistant in earlier chapters, its environment was simple: just text input and text output. Then we added tools, and suddenly its environment included calculators and search engines. We added memory, and now its environment includes conversation history. We added planning, and it gained the ability to coordinate multiple actions over time.

Each new capability expands the environment. This is natural and good, but it means we need to stay thoughtful about what we're adding and why. Every expansion of the environment is an expansion of what the agent can do, which means it's also an expansion of what could potentially go wrong.