Fund Fact Sheet Intelligence Copilot with Amazon Bedrock

Asset managers and wealth advisory teams face a recurring intelligence gap: critical product information is locked in dense PDF fact sheets, while teams are expected to provide quick, accurate, and defensible answers. idp.ai adds an AI layer for fund ingestion, enrichment, and conversational retrieval. Every answer is grounded in the source fact sheet and the user stays in control: it is a source-grounded retrieval copilot, not an AI analyst.

Multimodal Page Extraction: Each fact sheet page is converted to an image and read by multimodal models on Amazon Bedrock, capturing text, tables, and visual data. GCC Fund Enrichment: Context-aware field enrichment for GCC/UAE fund structures and Shari'a context produces structured JSON covering fund profile, holdings, allocations, risk/performance, and share class fields. Semantic Indexing: Content is embedded with Amazon Titan embeddings on Amazon Bedrock and indexed into Pinecone for high-relevance, fund-scoped retrieval. Source-Grounded Q&A: A RAG assistant answers plain-language questions, such as the top 5 holdings and risk rating, using only retrieved fact sheet content, and returns PDF page citations so users can verify every answer. Processed Document Registry: Documents already ingested are detected and their existing metadata is returned, avoiding re-ingestion cost on every app restart.