PinnedInArtificial Intelligence in Plain EnglishbyAli Elmalı·Jul 6Don’t Put Everything into RAG: Building a Safe Candidate Layer for Web SourcesInstead of feeding information found on the internet directly into RAG, first put it through the cache, review, and export stepsA response icon1A response icon1
PinnedAli Elmalı·Jan 13, 2024All articles at a glance — ( English)(17) English Episodes:A response icon10A response icon10
PinnedAli Elmalı·Jan 13, 2024All articles at a glance — (Turkish)(102) Turkish episodes:A response icon3A response icon3
InAI Advances BriefsbyAli Elmalı·Sep 17From Desktop Commander to MCP: How I Set Up My Own Workflow with ChatGPTI was using Desktop Commander with ChatGPT. I replaced it with my own MCP-based working layer on Local LLM. The reason isn’t just the cost…A response icon2A response icon2
InArtificial Intelligence in Plain EnglishbyAli Elmalı·Sep 16What Questions Would I Ask Before Deploying an Enterprise AI Platform into Production?When evaluating an enterprise AI platform, the key question is: Can it be seamlessly integrated into the existing architecture, security…A response icon1A response icon1
InArtificial Intelligence in Plain EnglishbyAli Elmalı·Sep 13Why Does the Security Model Change When You Enter AI Agent Production?An AI agent doesn’t just generate responses; it calls for a ride, reads files, executes commands, and sometimes makes changes to real…A response icon1A response icon1
InArtificial Intelligence in Plain EnglishbyAli Elmalı·Sep 13With Git and a Container Registry Already in Place, Why Might We Need a Layer Like KitOps?With “Jozu” and “KitOps,” rather than introducing a new tool, it is possible to transform the AI system into a DevSecOps architecture in…A response icon2A response icon2
InAI Advances BriefsbyAli Elmalı·Sep 10What Did I Learn Before Connecting FVG to the Decision Engine?My pilot study, conducted on six BIST stocks using 1,530 daily candlesticks and 378 FVG patterns, showed that data quality, measurement…A response icon2A response icon2
InBecome BetterbyAli Elmalı·Jul 1Before Scaling RAG: Ensuring Technical Terms Are SafeTransitioning to a local glossary guard, regression tests, and an agentic web ingestion design to reduce local LLM predictionsA response icon2A response icon2
InBecome BetterbyAli Elmalı·Jun 20Setting Up the RAG System Wasn’t Enough: This Time, I Proved That the Source Was Actually ProcessedHow did I use Batch Guard to resolve issues with files having the same name, incorrectly processed records, and source ID problems across…A response icon2A response icon2