Machine learning and LLM integration. Explore 22 articles in this category.
BlueRobin's Archives looks like a RAG app from the outside. Inside, it's a collection of specialist agents — triage, finance, health, tax, entity resolution, insight, and retrieval — each with its own job, coordinating over an event bus.
How BlueRobin combines multi-model vector search, Reciprocal Rank Fusion, knowledge graph traversal, and full-document context assembly to answer complex questions across your entire archive.
A walkthrough of how BlueRobin transforms any uploaded document — PDF, image, or Office file — into a fully analysed, encrypted, and semantically indexed record.
How BlueRobin extracts named entities from every document, resolves them into a canonical taxonomy, and stores a property graph that connects people, organisations, dates, and events across your entire archive.
How we use Model Context Protocol to create a closed-loop design system where agents verify implementation against Figma specs automatically.
Bridging the gap between design and development by using the Model Context Protocol (MCP) to automate UI generation from Figma to Blazor.
Implementing model policy management and token accounting so LLM features remain cost-aware and governable at platform scale.
A pragmatic analysis of the costs, performance, and complexity of running local LLMs versus cloud providers, and why we adopted a hybrid architecture.
What happens when agent graph node names collide with state keys, and how to design LangGraph flows that remain safe under change.
How to introduce an agent-based retrieval routing layer with health-aware fallback to keep retrieval reliable in production.
A production blueprint for combining graph traversal and vector search to improve recall and precision in document retrieval.
How to evolve entity extraction by combining deterministic and LLM-based providers with confidence-aware ensembling.
How to combine multiple retrieval strategies with reciprocal rank fusion to produce stable, high-quality contexts for LLM responses.
Build intelligent AI agents using Microsoft Semantic Kernel with tool calling, memory, and multi-agent coordination in .NET.
How to extract entities (NER) from documents using LLMs and model them effectively in FalkorDB for knowledge graphs.
Extract named entities from documents using spaCy NLP service for building knowledge graphs and improving search relevance.
Techniques for rewriting user queries (HyDE, Expansion) and reranking results to boost retrieval accuracy in Retrieval-Augmented Generation.
Implement a complete Retrieval-Augmented Generation pipeline that combines semantic search with local LLM inference for intelligent document Q&A.
Build a powerful hybrid search system that combines vector embeddings with traditional keyword search for comprehensive document retrieval.
Create a production-ready embedding pipeline in .NET that generates vector embeddings with Ollama and stores them in Qdrant for semantic search.
Implement effective text chunking strategies for RAG pipelines with semantic boundaries, overlap, and metadata preservation.
Learn how to integrate Docling, an AI-powered document understanding library, into your .NET application for high-quality OCR with layout preservation.