Date: 2026-04-27 | Newsletter Key Summary DeepSeek’s V4 models landed this week: 1.6 trillion parameters, 1 million token context window, performance rivalling closed-source frontier models, trained for roughly $5.6 million, and released under a true MIT license with no user…
Date: 2026-04-25 | Newsletter Key Summary Two weeks before Grab hosted GrabX 2026 in Jakarta — announcing 13 AI-powered features, an “Intelligence Layer,” and an autonomous delivery robot named Carri under the banner of “responsible AI” — Zenta had a…
Date: 2026-04-24 | Newsletter Key Summary A 2026 report on AI citations revealed that Reddit is the #1 most-cited source across major AI platforms, appearing in roughly 68% of AI-generated answers. On Perplexity alone, ~24% of all citations come from…
Date: 2026-04-23 | Newsletter Key Summary A Reddit post captures a familiar pain: “I subscribed to 6 AI tools this quarter and I’m actually getting less done.” A reply echoes it: “$200 a month, and the worst part isn’t the…
Date: 2026-04-21 | Newsletter Key Summary Everywhere in Asia, business owners are debating whether to allow employees to use AI. Zenta reframes the debate: banning doesn’t work (a global study of 32,000 workers found 40% will violate company policy to…
Date: 2026-04-20 | Newsletter Key Summary Two major consulting reports landed the same week: PwC surveyed 1,217 executives across 25 sectors and found the top 20% of companies capture 74% of AI’s economic value, generating 7.2x more AI-driven revenue than…
Date: 2026-04-18 | Newsletter Key Summary An MIT study found that 95% of enterprise AI pilots delivered zero measurable ROI — not low ROI, zero. A CIO.com survey adds texture: 40% of US businesses are still getting bulk AI value…
Date: 2026-04-17 | Newsletter Key Summary Uber’s CTO admitted publicly that AI costs “blew away” projected budgets after the company gamified Claude Code adoption with internal leaderboards — driving 11% of live back-end updates to be AI-written in three months,…
Date: 2026-04-16 | Newsletter Key Summary Three moves in 24 hours — and Zenta reads them as one game. OpenAI launched GPT-5.4 Cyber, a cybersecurity-tuned model with relaxed guardrails for binary reverse-engineering, positioned as a counter-punch to Anthropic’s Mythos. Anthropic,…
Date: 2026-04-15 | Newsletter Key Summary Three seemingly unrelated events in 24 hours share a single thread. Anthropic announced Project Glasswing — a consortium of AWS, Apple, Google, Microsoft, NVIDIA, and major financial institutions — to defend against its own…
