Zhemal Khamidun
Head of AI at Alpina Digital, CPO of AlpinaGPT (~30-person engineering & AI team). Founder of Vibe Academy & Hamidun Consulting, author of the AI channel «Готовим ИИшницу» (@jemal_hamidun).

Svoi.ru cuts test preparation time by 70% with AI agents
Svoi.ru's team showed how AI agents can automate requirements analysis and test documentation prep, taking routine analysis off QA's plate and cutting that stage by…

Kodik explains why public LLM benchmarks mislead
Kodik dissected the weak spots of popular LLM tests and argues that an AI code editor needs its own benchmark, not flattering percentages from someone else's tables.

How Google DeepMind and Rivals Are Changing Music: Five AI Services for Generating Tracks
A roundup of five AI services shows how text-to-music generation has stopped being a toy and become a working tool for songwriters, brands and independent producers.

OpenAI, Anthropic and Gemini: How Inference Caching Cuts LLM Cost and Latency

Habr walkthrough shows how to train a mini-LLM in C# using ILGPU and integrated AMD graphics

Anthropic unveils Claude Mythos Preview with a 244-page system card instead of a release

Anthropic CEO Talks Mythos Access With White House Amid Pentagon Dispute

OpenAI and Anthropic change how LLM costs are calculated: in 2026, the price of a task is what matters
OpenAI and Anthropic are changing LLM pricing rules: in 2026, businesses can no longer just watch the price per token. They will have to calculate the full cost of…

Allbirds Sells Its Shoe Business and Pivots to AI Cloud as NewBird AI
Allbirds is selling its footwear business for $39 million, renaming itself NewBird AI, and betting on renting out GPU capacity to the AI market.

Anthropic heads to the White House: Mythos escalates the debate over AI and cyber risk

Canva builds its design engine into Claude through a partnership with Anthropic

Intel pits Wildcat Lake chips against MacBook Neo, betting on local AI

Anthropic CEO to meet White House chief of staff as US government seeks access to Mythos model

Claude Code and Codex: How Three Markdown Files Cut Token Waste
Claude Code and Codex often burn most of their context re-navigating projects. A hierarchy of a global map, CLAUDE.md and past-session memory fixes that.

LM Studio and Qwen: How Local LLMs Handle Coding on a MacBook M4 Pro
The author tested local Qwen, Gemma and other coding models through LM Studio on a MacBook M4 Pro: they are viable in chat, but noticeably trail the cloud in agent…









