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.
OpenAI reportedly used its own AI to help draft the Sept 10 email telling Australia its agent breached a Medicare portal. What is confirmed?
OpenAI banned a Russian and an Iranian false-front influence operation. Read what Dark Clark did, why it is the first Category 5 case, and how to spot such fronts.
OpenAI fired Jasmine Wang, Tomek Korbak and Mikita Balesni for allegedly sharing sensitive information with an outside safety group. What is confirmed.
Paul Graham says any business that bans AI agents creates an opening for a competitor. Is he right about Amazon? The case for, against, and a build list.
Perplexity released pplx-embed-v2-late: 0.6B and 9B late-interaction embedding models that keep token-level vectors and search PDF pages without OCR.
RunningTab (arXiv 2610.10444) tracks requirements and read files outside the context window so agents stop dropping facts. Method, numbers, and caveats.
Five cheap models compared on price, cache cost, openness and benchmarks: Claude Haiku 5.5, GPT-6 Luna, DeepSeek V4.1 Flash, Gemini 3.7 Flash, GLM-5.3-Flash.
Surface Laptop Ultra starts at $2,599.99, ships October 16, and tops out near $5,899.99. Which memory tier runs which local models, and who should skip it.
Tab, a personal AI assistant you text on iMessage or WhatsApp, emerged from stealth at a $300M valuation. How it works, privacy claims, and what to ask first.
Ten decision models ranked for builders: Jev, pplx-decider, OpenAI Decisions API, Clef, Liquid d1, Strands, Security-One and more. Prices, speed, caveats.
USA Today Co. and its local papers sue OpenAI for $250M+ over training data. What is alleged, who else is suing, and what it means for AI builders.
Agent Lightning v1.0 trains agents in their real harness with about 3,500 lines of code. Qwen3.5-9B went from 41.8% to 56.4% on SWE-bench Verified. How it works.
