Environmental Assessments using RAG LLM and Agentic Workflows
Queriable data from 25+ documents for time savings of 200+ hours per assessment
TYNI Problems
25+ Documents with handwritten text and images
200+ hours condensing and summarizing information
Up to 1500 pages of manual data entry
TYNI Solutions
Data extraction with pymupdf
+
Recursive chunking with langchain
+
Embedding with openapi
+
Vector DB with Milvus
+
Route LLM for decision making and Reactagent to run Gemini and Llama models
TYNI Results
Queriable data from all documents
Assessment produced in less than 30 minutes
Minimal manual adjustment and less chance for human error
