Abstract
The present disclosure relates to the field of Artificial Intelligence (AI), in particular to agentic AI framework for semantic transaction data compression and retrieval. The disclosed system comprises a plurality of AI agents, including an ingest agent, a selector agent, an encoder agent, a retriever agent and a governance agent, collectively configured to process large volumes of transaction records generated by electronic payment networks. In operation, transaction records are ingested, normalized, and analyzed to dynamically determine an appropriate compression policy based on storage efficiency, semantic fidelity and retrieval latency. Compressed transaction data is stored in an object storage system along with semantic digests comprising textual summaries and embedding representations, and metadata. Upon receipt of a query or decompression request, the system reconstructs one or more transaction records using semantic search and performs automated verification using embedding similarity checks and a generative LLM-based verification loop to preserve business-critical transaction meaning.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
SHANGLE, PALAKH, "AGENTIC AI FRAMEWORK FOR SEMANTIC TRANSACTION DATA COMPRESSION AND RETRIEVAL", Technical Disclosure Commons, (February 03, 2026)
https://www.tdcommons.org/dpubs_series/9274