How to interact using different LLMs?

Different large language models excel at different functions. Is it currently possible to integrate various models to handle different tasks, with agents able to interact with one another, and have this achieved with only simple configuration?