Latest posts
-
How to Build an Evaluation Set — Proving Your RAG, Text-to-SQL, and Agent Improvements with Numbers
You changed the prompt, RAG, and SQL flows — but did things really improve? A practical guide to building an evaluation set: sourcing questions, defining answers, sizing, grading, per-stage measurement, and upkeep. -
How to Serve Internal Database Data to an LLM — Tool Calls, Text-to-SQL, and Semantic Layers
Document RAG cannot answer questions about numbers in your internal database. Learn how to choose between tool calls, Text-to-SQL, and a semantic layer, plus the security and evaluation rules that matter. -
Prompting, RAG, or Fine-Tuning? How to Choose Based on the Failure You See
When should you use prompt engineering, RAG, or fine-tuning? Follow a team building an internal AI assistant and learn the failure symptoms, decision order, and checklist that tell you which to pick. -
RAG Fine-Tuning with Synthetic Queries and Hard Negatives: A Research Survey and Practical Recipe
Summary — Part 1 built a RAG pipeline with chunking, hybrid search and reranking, and Part 2 rebuilt it in Langflow. If retrieval still misses because of internal jargon and document structure, the remaining lever is adapting the retrieval model itself to your data. This post surveys the key ingredient for RAG fine-tuning: synthetic questions […] -
Rebuilding the Same RAG Pipeline in Langflow — What Low-Code Can (and Can’t) Do
Summary — In Part 1, I built an internal-manual RAG pipeline (chunking → hybrid search → reranking) in LangChain code. Here I rebuilt the same pipeline in Langflow. Bottom line: the pipeline’s skeleton maps cleanly to low-code components, but fine-grained logic like dense+sparse hybrid queries and RRF fusion still has to drop down into a […] -
Building a RAG Agent for Internal Manuals: Chunking to Reranking
Summary — A code-level walkthrough of turning internal manuals stored in Confluence into an agent that answers questions with real evidence. We split it into two flows — ingestion (collect, chunk, index) and query (hybrid search, rerank, generate) — and cover the actual libraries, APIs, and recommended parameters used at each step. → See the […] -
AI’s Next Advantage Is the Meaning of Data
Moving beyond productivity gains to customer value — the semantic layer and Ontology What This Article Argues Organizations are already deploying AI to draft documents, search for information, write code, and handle customer inquiries faster. Real productivity gains from this are already being documented. But processing work faster doesn’t, by itself, create new value for […] -
Why Enterprise AI Stalls at the Chatbot Stage
Who This Article Is For The chatbot pilot went smoothly. Then, right before a company-wide rollout, it just stopped. This is for anyone who’s run an internal chatbot pilot and hit a wall at the actual deployment stage — or a practitioner who just got asked, “let’s build a chatbot on our company data too.” […] -
AI Agent Concepts, Hands-On: Build a Budget Summary App
Summary — Reading about AI agents, prompt/context engineering, harness engineering, and agentic engineering only gets you so far without hands-on practice. This article builds a simple “budget summary app” together, walking through all four concepts step by step. Instead of real screenshots, each step includes the exact example prompts you’d type and example responses AI […] -
Agentic Engineering and the PH-AH Loop: Not Vibe Coding
Summary — Agentic engineering is a collaboration style where AI produces the work, but a person reviews and confirms it at every planning and execution step. This repeating structure is called the PH-AH loop (Plan-Human, Action-Human). It sits at the exact opposite end of the spectrum from vibe coding — “using AI-generated code without ever […]