MCP + RAG: How AI Turns Data into Clear Answers, Tables and Charts
How to combine MCP and RAG for document search, Telegram conversation analysis, financial data and telemetry, with practical examples.
MCP for Websites: Audits, Product Selection and Business Data for AI Clients
A practical guide to website MCP: data audits, product selection, support, analytics, security and real ifreework.com scenarios.
“Let’s hire AI bots and make support cheaper.” What could go wrong?
A simple balance transfer took about five hours. Here is why chat history cannot replace task state and what backend architecture AI support needs.
Jev vs. Five LLMs: Selection Quality, Latency, and Cost for Building RAG Systems
Which model fits each RAG task: bulk filtering, fast review, or strict source selection? We compare Jev and five LLMs on 200 projects and show how aggregators such as apixly.ai help run the evaluation.
TypeSafe and Jev in Laravel: Lessons from Evaluating 4,344 GitHub Projects
A practical IdeaRadar experiment: integrating TypeSafe/Jev with Laravel, evaluating 4,344 repositories, and examining why fast, structured responses do not guarantee sound selection.
Running a Local LLM in Production: Integrating Ollama with Laravel
How we added the Qwen2.5 7B via Ollama to the production pipeline, tested the CPU inference and came up with a hybrid scheme: local model for primary analysis, DeepSeek for complex candidates.
Wire Room
How the AI news pipeline Wire Room works: RSS collection, rewriting, search for duplicates and publication on the site and in channels. Separately, I show the scenario of sending the news by the editor through the Telegram bot.
Book Processing Pipeline
How photos of books turned into cards on Ozon, Wildberries and Yandex Market. A look at Laravel system with data recognition, pricing, publication and synchronization of balances.
Building “Electronic Brain”: an AI Knowledge Base with GPT, Email, Voice Notes and RAG
How an AI knowledge base works, combining emails, documents and voice notes. I analyze the processing of sources, indexing of RAG database and searching for information through a dialogue with GPT.
Building an Intelligent News Pipeline
Development of the news pipeline on Laravel: collection of materials, processing by neural networks and preparation for publication. I talk about background tasks, status control, API errors and editor interface.