TechTouch Asia

- Over 1,500 files uploaded across multiple formats - Nearly 10,000 pages ingested - Over 90% reduction in document screening time Staff from the Singapore Management University’s (SMU’s) admissions office used to spend hours scrolling, copying, pasting, and double-checking student applications to identify qualified candidates. Wanting to speed up these time-consuming manual processes, it turned to automation. But instead of alleviating the situation, new problems emerged. Automations frequently broke down due to minor user interface changes, leading to high maintenance effort and frequent disruptions. The problem with typical AI agents is that while they can understand and process simple templated documents, they struggle to read documents that come in varying formats, or which contains elements like embedded tables, graphs, and inferred values. What SMU needed was intelligent document processing, where agents can operate effectively in unstructured environments, tackle complex prob...