RAG and retrieval — page 4

Local-first hybrid RAG: traditional vs Jev-style (System One) pipelines over your own documents — with a built-in benchmark lab (6 scenarios, 48 questions, independent LLM judge). Hybrid won +8.4pp correctness at identical cost.

Simple personal profolio plus chat assistance with RAG and Jev

Run the new wave of decision models (Laya, Decider, Kev, Jev) side by side on your Mac. Structured input in, calibrated probabilities out, with honest accuracy, calibration, latency and memory numbers.

Use Jev instead of an LLM for RAG decisions: relevance grading, query routing, hallucination checks and LLM-as-a-judge evals. Faster, cheaper, measured first.

Semantic emoji search - keyword + embedding recall, reranked by TypeSafe Jev. Built with TanStack Start, TanStack Query, and Emoji Mart.

Version-aware evidence decisions for RAG assistants using Jev, with inspectable routing, bounded recovery, and reproducible evaluation.

Two-stage session and passage retrieval with optional Jev reranking

RAG evaluation with typed decisions: sentence-level hallucination verdicts with offsets and citations, passage and answer relevance, in one Jev call per answer. Benchmarked against Ragas, DeepEval, an LLM judge and HHEM.

kblam: the knowledge base for LLM-assisted mereology. it's a flat-file research/knowledge storage system with enforced organization (using heuristics, semantic similarity, and Jev evaluation) to prevent document drift, internal contradictions, incorrect data gleaned from reading superseded sections out of context, and other LLM bullshit. not done

A precision-first RAG orchestrator: filters before it generates, verifies before it answers.

Jarvis — assistente de IA pessoal em Google Apps Script: agente ReAct (Gemini), RAG no Drive, voz no Android e julgamentos tipados (TypeSafe/JEV)

Open-source typed decision service: in-process embedding engine, JS/Python SDKs, self-hostable

Multi-tenant RAG assistant for the SaleQue CRM: cited answers on PostgreSQL and pgvector with Row Level Security, Claude and OpenAI behind one adapter, Jev lead scoring. Reference build.

Adaptive RAG Router using TypeSafe Jev, with budget constrained evaluation against rule based and LLM routers.

OpenClaw, powered by Jev. Jev-powered memory decisions with your existing LLM as fallback. Built with Jevify.

Pluggable, observable RAG MCP server — hybrid retrieval (BM25+dense+RRF), BEIR-benchmarked rerankers (BGE / TypeSafe Jev / LLM), Streamlit dashboard + Langfuse, multimodal ingestion

An open-source, AI-powered agentic RAG demo that routes questions to documents, OpenAI docs, or the web.

AI Engineering knowledge base & Obsidian automated extractor (TypeSafe Jev + DeepSeek)

Jev decision toolkit for DeepSeek Harness: 22 named typed judgments (privacy scan, change-scope, memory triage, batch triage) on one hardened, budgeted, ledger-backed transport. Advisory only.

A playground for TypeSafe AI's Jev (System One model), built around five real production workflows: support triage, RAG relevance gating, agent action firewall, inline moderation, and CI eval judging.

Drop-in rerank API served by Jev: speaks the Cohere, Jina and Voyage rerank protocols, so any RAG stack switches by changing a URL.

Jevtrieval is a retrieval-augmented generation demo. It retrieves documents from Qdrant, scores their relevance with Jev, and generates an answer from the selected documents.

Comprehensive poker game theory knowledge base, instruction-tuning dataset, and TypeSafe AI (Jev) evaluation benchmark for Limit & No-Limit Texas Hold'em.

Question classification experiment using TypeSafe AI Jev for QA/RAG routing.