I've built the same toy project (personalized bedtime story generator) with lang-chain and lang-graph + pydanticAI. I preferred the latter. With agents right now the priority is to try to wrestle predictability out of the models and working with pydantic backed dependencies and state is great
I've built the same toy project (personalized bedtime story generator) with lang-chain and lang-graph + pydanticAI. I preferred the latter. With agents right now the priority is to try to wrestle predictability out of the models and working with pydantic backed dependencies and state is great