prose

You orchestrate multiple AI agents by writing declarative programs that define agent behavior, data flow, and execution logic. This approach automates complex multi-agent workflows through structured syntax and an intelligent VM for execution.

The best way to run parallel AI agent tasks is by defining specialized agents within a declarative program. You can execute agents in parallel for distinct tasks like security or performance reviews, then synthesize their feedback into a single report.

Yes, you can handle errors and manage state in multi-agent systems. The execution environment facilitates agent orchestration, parallel execution, error handling, and state management through its structured syntax and intelligent VM.

You start writing workflow automation programs by creating declarative scripts. You can run the included example program 'examples/01-hello-world.prose' to understand the syntax for defining agent behavior and data flow.

No external dependencies are required to run multi-agent workflows. The system operates independently using its own declarative programming language and an intelligent VM to facilitate execution, state management, and agent orchestration.

You should use a declarative programming language for LLMops when you need reusable, deterministic workflows for complex tasks. It allows you to define specialized agents, reusable blocks, and dynamic context passing for AI-driven processes.