At Remote, we help companies hire, manage, and pay talent anywhere in the world. When onboarding customers, getting their existing HR and payroll data into our platform quickly and accurately is a critical part of the onboarding journey.

To solve this challenge at scale, our engineering team built a Code Execution Agent using LangChain and LangGraph. Instead of asking an LLM to process data directly (which leads to hallucinations and context limits), our agent “thinks” by planning the migration and “acts” by writing and running deterministic Python code in a secure sandbox.

This hybrid approach has transformed a manual process that took days into an automated workflow that finishes in hours.

We recently partnered with the LangChain team to document exactly how we built this architecture, why we chose LangGraph for orchestration, and the lessons we learned about keeping data processing outside the LLM context window.

Read the full case study on the LangChain Blog →