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.
Garry Tan's Aug 24 thread: systems of record need AI harnesses or agents replace them. DB/API/ACL stay; hallucinations are workflow problems.
Grok Build --agent-budget sets cumulative subagent caps (1–1,024) for multi-agent workflows — landed in v1.0.7, not v1.0.9 per public changelogs.
Higgsfield Blender add-on (Aug 25, 2026): prompt blockouts, animate cameras, reblock shots in seconds. MCP + Supercomputer paths, limits, and comparisons.
Lars Faye argues AI code generation skips the friction that builds developer taste. JetBrains and UPenn studies back the worry — but Socratic use can help.
McKinsey Aug 2026: 80% productivity, 37% EBIT impact, 32% build in-house instead of buying SaaS, 40% of large enterprises scaling agents — practitioner read.
Namespace unboxed MacBooks for server racks — 2.8M views asking why not Mac minis. Yes, Macs work as CI servers; here is why a dev cloud might pick laptops.
OpenAI DevDay is Sept 29 at 10 a.m. PT — no official product list, but September already shipped Agents API and GPT-Live-1. Here is what the keynote is likely to add: ChatGPT o, Pro Max, Work/Codex, hardware teasers.
Paul Graham's Aug 25 essay says YC wants builders with a habit of side projects — not entrepreneurship majors. Harvard vs Yale data, reading period, and why business plan contests mislead.
Perplexity Computer GPT-5.6 Sol costs less after OpenAI's official 20% API cut — not a separate Perplexity discount. Gateway math and Computer billing explained.
Perplexity demoed a fully local Computer on NVIDIA DGX Spark Aug 25, 2026 — orchestrator + subagent + harness on-device, PPLX 27B and Qwen 3.8 27B local, cloud routing user-gated. Requires Pro/Max + DGX Spark hardware.
Polymarket said Porsche signed a $1.5B AI deal with an Indian firm. The real story: TCS buys MHP for €320M plus a five-year AI partnership — not a hype tweet.
Qwen3.8-Flash-Next 125B-A6B MoE drops Aug 26 on ModelScope/HF. Leaked specs, Qwen4 architecture teaser, 128GB Mac fit, FreeToken offloading, vs Qwen3.8-27B.
