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.
Grok Bot 0.70.2 gives each bot a mail.grokbot.com inbox, approvals for irreversible actions and X search without login. What is confirmed, what is not.
How to build an always-on AI agent: server-side loop, wake triggers, sandbox VM, memory files, durable workflows and quiet-by-default alerts, step by step.
IBM Granite Switch 4.1 (8B, 30B) bundles chat, decision, Guardian safety and speech-to-text in one Apache 2.0 model. Setup, scores and what is unverified.
ICE posted an RFI for 24/7 AI that transcribes detainee calls, builds voice profiles and flags camera events. What it says, what is unknown, and the risks.
Into the Unknown: nine frontier agents wrote 135 follow-up papers for DOE lab problems; none passed all conditions and 114 failed the integrity check.
Jensen Huang told CNBC in London that Nvidia must work with Chinese AI firms on AI safety. What he said, the policy backdrop, and what is not yet clear.
Jev 1.14.0 adds opt-in images, a box type and 262k context at the same $0.042/M price. What TypeSafe confirms versus what headlines claim.
Krea 3 edits at native 2K in under 8 seconds for $0.06 per image. Krea's own comparison with Nano Banana Pro and GPT Image 2, access and limits.
Liquid AI d1-ultra: multimodal decision model, $0.042 per million input tokens, claims +8 points on a decision index. Claimed vs verified.
Folha tests found Meta AI on WhatsApp refused Lula posters but made Flávio Bolsonaro ones. Meta, the TSE and what is verified.
Muse Spark Open is Meta's 816B open-weight MoE model (42B active, 1M context). Specs, hardware needs, licence and how to run it.
METR proposes embedded assessments of alignment, control and AI R&D speed-up at frontier labs: what is proposed, what it builds on, and its limits.
