Leidos Used AI Agents To Onboard Agency Workflows 40% Faster
The Common Question
Can agentic AI handle multiple government document types without rebuilding the pipeline?
The Lesson in This Case
Leidos decomposed a monolithic document processing pipeline inside its ManagedX platform into discrete AI agents handling OCR, classification, field extraction, validation, and search indexing, so adding a new government agency or document type no longer requires modifying the core system. The multi-agent architecture, built on the Strands Agents SDK and deployed on AWS GovCloud (US), assigns lighter models to classification tasks and more capable models to validation, keeping costs in check through separate throughput quotas per model. Senior leaders evaluating document-heavy workflows should take note that architectural flexibility is what allowed Leidos to scale across healthcare claims, legal discovery, and financial documents without rebuilding from scratch each time.
Originally reported by Amazon Web Services (AWS). Read the full story here.