In the field of artificial intelligence, Modular Reasoning, Knowledge, and Language (MRKL) systems aim to solve complex problems by integrating specialized modules. This presentation introduces an experiment conducted using watsonx.ai's foundation models, combined with a LangChain/ChainLit implementation. The focus of our experiment is a target pricing model that processes financial forecast queries through a multi-stage MRKL system. An initial MRKL agent receives user queries and uses a Large Language Model (LLM) for primary reasoning. This LLM accesses data from various sources, including the internet and a purposely curated datastore, facilitated by LangChain. The information collected is then transferred to another MRKL agent, which focuses on data synthesis and generating responses.