Expert profile
Founder & AI Product Leader
Yash Thakker is a Generative AI expert with over 12 years of experience in product leadership and technical strategy. As the founder of explainx.ai, he has taught over 300,000 learners and built AI platforms serving millions of users globally. He specializes in Agentic AI, Multimodal RAG, and the intersection of LLMs with consumer hardware.
Tim Cook's last day as Apple CEO was August 31, 2026. John Ternus, Apple's hardware chief, takes over September 1 — here's what a hardware-background CEO signals for Siri, Apple Intelligence, and the Gemini deal.
Tsinghua researchers broke the 41-year Dijkstra SSSP record with O(m log^{2/3} n) — recursive partial ordering instead of sorting every node by distance.
Vercel's design.md encodes colors, typography, and spacing for AI pages — spec-driven rendering with test loops, but community says slop persists in code.
Waymo lists 14 US robotaxi cities in September 2026. See which are public, Uber-partnered, or still on waitlist — plus July's four-city expansion.
Grok's Outlook connector gets Mail.ReadWrite and Mail.Send at first sign-in — full send access unlike Gmail's read-only default. Setup and IT checklist.
GLM-5.3's "50% coding boost" traces to one internal benchmark, not across-the-board gains. Here's the real numbers, the new license, and how it compares to open-weight rivals.
OpenAI says ChatGPT Ads reached a $1B annualized revenue run rate in under 200 days, with self-serve buying now live in India, Europe, MENA, and North America. What the number means and what it doesn't.
OpenAI is opening ChatGPT Ads self-serve buying in India as part of a four-region expansion. What Indian advertisers and marketers should know about inventory, the Go plan, and DPDP Act targeting limits.
OpenAI cut giant ChatGPT thread load time 94% in tests — 741 turns, 231 MB, 27.6s → 1.7s. Client-side history, not model speed.
Simon Willison vibe-coded a codex.tools reference snapshot — 232 tool interfaces, 44 skills, ~615k chars of skill source. explainx.ai maps what builders learn from it vs asking the agent, and why control-browser is a skill.
Code-as-World represents physical scenes as executable code so VLMs can reason quantitatively. The Aug 26 paper hit #1 on Hugging Face Papers with QuantiPhy SOTA.
Tencent Cloud CubeSandbox v0.7 adds cross-node pause/resume, CubeOps control-plane split, and faster networking — 11.5k GitHub stars, E2B SDK compatible, Apache 2.0.
