TLDR Tech / Dev
How Grab is Using AI Agents to Boost Team Productivity
Grab's data engineering team built a multi-agent system to automate investigation work that was consuming two full days a week of senior engineers' time. Supporting over 15,000 tables and 1,000 monthly users, the architecture splits work by risk profile. A read-only investigation pathway uses Classifier, Data, Code Search, On-call, and Summarizer agents orchestrated via LangGraph, while a more cautious write pathway employs a single Enhancement Agent gated by human review.
Read the full story