Habr AI
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications
Sber Develops GigaChain Without Forking LangChain, Betting on a Compatibility Package
Sber explained how it develops GigaChain around GigaChat: the team rejected a heavy LangChain fork, built a compatibility package, and is creating an open AI agent ecosystem.
Timeweb Cloud explains how edge detection filters work: from Roberts to Canny detector
Timeweb Cloud released a detailed guide on edge detection in images: it covers derivatives, Roberts, Prewitt, Sobel, Laplacian, Canny operators, and the Wallace filter.
Peter Publishing House released a book on conversational AI and chatbots that work
Peter Publishing House presented a book on effective conversational AI: it focuses on LLM-based chatbot creation, dialogue design, and real user experience.
SpeShu.AI explained how to find work in AI amid growing job vacancies in Russia
SpeShu.AI highlighted a market paradox: the number of AI-skill vacancies in Russia reached 16.5k per quarter, but high competition and hundreds of applications complicate hiring.
Anthropic's Opus 4.7 consumes more tokens: hidden inflation at unchanged prices
In Anthropic's Opus 4.7, the same text may require significantly more tokens than in Opus 4.6 — with prices unchanged, this is a subtle but real increase in API costs.
Alfa-Bank Showed How Skeptics Deployed an Internal HR Service to Production with GLM-5
Alfa-Bank described an experiment with GLM-5: a team of team lead, analyst, and product manager shipped an HR service to production in just a few weeks, but the main bottleneck turned out not to be AI, but Git.
How a YADRO Developer Configures Claude Code with Spec-Driven Development
A YADRO architect has been managing AI agents using SDD methodology for nine months and explains how to configure Claude Code to switch models between projects without reloading.
How Just AI Helped a Bank Break Through Automation Ceiling with LLM-Agents
Hundreds of NLU-scenario branches — yet the automation rate still stalled. Just AI migrated the bank's support to LLM-agents and added a judge-agent against hallucinations.
10 RAG approaches that actually work in production: from basic to GraphRAG
A practical breakdown of 10 RAG techniques — from naive RAG and hybrid search to GraphRAG and Self-RAG — with guidance on when to apply each in real-world projects.
SENSE: why the LLM market is shifting models from service to paid access to thinking
The SENSE author proposes viewing LLMs not as another SaaS offering, but as infrastructure: tokens become the unit of access to thinking, price and security emerge as the primary competitive arena.
Sber: Yago knowledge graph barely helped search, while LightRAG added 12 percentage points of accuracy
Sber team compared two approaches to improving search: the ready-made Yago graph barely changed results, while LightRAG built on their own documents delivered 12 percentage points of accuracy gains.
Hugging Face and ModelAudit: research revealed the limits of built-in ML model protection
An experiment on Hugging Face models showed that ModelAudit detects additional dangerous signals, but often generates noise, and the accuracy of findings strongly depends on rule versions and manual triage.
Local Vision for z.ai GLM-5.1: 8B Model Closes 70% of the Gap to the Frontier
An author demonstrated how to add local vision to a low-cost coding agent based on z.ai GLM-5.1 via an MCP server and qwen3-vl:8b, enabling it to verify the interface from screenshots.
Federal Treasury Launches AI Monitoring of National Project Construction Sites Using QMonitoring
Federal Treasury has integrated QMonitoring for 24/7 monitoring of national project construction sites: AI-powered video analysis detects risks, failures, and schedule delays.
Aurora Proposes a Manifesto of Sustainable AI — with Memory, Identity, and Development
A manifesto on "sustainable AI" was published on Habr AI: the author proposes evaluating systems not by benchmarks, but by their ability to maintain identity, change, and reproduce over time.
SENAR introduces quality gates for AI development: how specifications and metrics reduce errors
A new part of the SENAR series was published on Habr: the author shows how mandatory specifications, input and output gates, and metrics like FPSR and MIR reduce the number of bugs and manual edits.
OpenWebUI breaks down how to connect System Prompt, Skills and MCP Tools for link validation
OpenWebUI's tech support bot example demonstrates why separating System Prompt, Skills, and MCP Tools is essential when you need not just to generate links, but verify they actually work.
Anthropic Changed Claude Opus 4.7's Character—And Some Developers Saw a Regression
After releasing Claude Opus 4.7, Anthropic promised a more direct and precise model, but some developers encountered excessive arguing, increased token consumption, and broken legacy prompts.
How five documents can break a RAG system and turn a knowledge base into an attack vector
RAG reduces hallucinations, but trusted context from the knowledge base itself becomes an attack point: a few poisoned documents are enough to skew model responses and bypass protections.
Habr AI reveals how a single open port in an LLM platform exposed API keys and infrastructure
A security audit team testing a RAG service with tens of thousands of users didn't break through complex attack chains, but obtained access to API keys and infrastructure via a single open port.
Anthropic Claude Won't Save You From Bad Requirements: Why AI Delivers Convincing But Weak Results
Using Claude as an example, the author demonstrates that AI often accelerates the delivery of plausible answers rather than useful results when a task is not broken down into input, stages, output, and acceptance criteri
How a Russian Developer Spent Days Launching Gemini—and What Finally Worked
A Russian programmer spent several days launching Gemini from Russia via VPN, tried dozens of approaches, and documented each step—what didn't work and what ultimately helped.
Product Graph and Agent Memory: Why AI Doesn't Save Products Without Knowledge Structure
An analysis of Product Graph explains why even powerful AI agents are useless without shared product memory and how interconnected knowledge structures reduce team chaos.
Laboratory Over Six Years: From USB Drives and Notebooks to AI That Finds Hidden Equipment Defects
A story about how one laboratory spent six years building digital infrastructure — and ultimately developed AI that reads deposition logs and detects hidden equipment defects in advance.
How AI Transformed Diary Research: Three Compromises We Left Behind
A UX research team explained how AI made it possible to abandon compromises on sample size, duration, and analysis depth—without sacrificing methodological value.
Marcin Moskala Audited GeminiAI: What the Code Review Revealed About Coroutines and Android Architecture
The author of the open-source GeminiAI client shared how his project passed Marcin Moskala's audit and why coroutine errors became the key test of Android architecture maturity.
Anthropic: Claude Code source leak revealed complex agentic architecture
A leaked sourcemap from Claude Code showed that Anthropic's product has long evolved from a 'chat CLI' into a platform with sub-agents, memory, remote modes, and strict shell security.
Sber explained why business needs an AI Overlay layer instead of restructuring departments
Sber showed why targeted implementation of generative AI rarely brings profit, and proposed an alternative — a horizontal AI Overlay layer on top of processes, data, and corporate systems.
Saiga Llama 3 8B on 10 GB VRAM: How Habr Achieved 93% Accuracy on War and Peace
Habr AI demonstrated how to run Saiga Llama 3 8B on 10 GB VRAM, compress two volumes of War and Peace into a summary, and reduce hallucinations through strict prompting and post-processing.
4 Non-Technical Founder Patterns That Ground Startups
A Habr developer who worked with several non-technical founders describes four patterns that prevent startups from launching—and how AI makes these patterns even more dangerous.