AI News · 2026-08-07

AI News · 2026-08-07
💡

Jason Says

Today's biggest signal isn't a product update — it's four DeepMind legends leaving simultaneously. When Jeff Dean, Sanjay, Oriol, and Quoc all walk out the door, you have to ask: how much of Google AI's moat was the people, not the infrastructure? The ripple effects are just beginning.

🛠️
AI ToolsTechCrunch AI / OpenAI Blog

ChatGPT Free Gets Unlimited Text Chats, GPT-5.6 Sol Upgraded

OpenAI is giving ChatGPT free and Go users unlimited text chats, plus a new 'think' button for complex queries. GPT-5.6 Sol also gets accuracy and consistency improvements. This signals OpenAI doubling down on free-tier growth as a strategic moat, putting pressure on competing AI assistants.

🛠️
AI ToolsTechCrunch AI

OpenAI AI Smart Speaker Reportedly Priced at $300–$400

OpenAI's upcoming AI smart speaker is reportedly priced at $300–$400, far above legacy smart speakers. This premium positioning signals OpenAI's bet that consumers will pay a premium for frontier AI hardware — a key signal of its broader consumer device strategy.

💰
MonetizationTechCrunch AI

Naïve Raises $28.5M to Automate Company Setup and Operations

Naïve raised $28.5M to take vibe-coding further — automating not just software, but the grunt work of running a company: registration, compliance, and operations. A compelling signal for solo founders and indie developers that AI can now handle company infrastructure, not just code.

🔥
SkillsGitHub Trending

Tencent Open-Sources TencentDB Agent Memory: Team-Level AI Memory Hub

Tencent Cloud open-sourced TencentDB Agent Memory, a team-level memory hub that converts conversations, docs, and code into four reusable asset types — shareable across agents and frameworks. Directly addresses the core pain point of collective amnesia in multi-agent collaboration.

📚
AI PapersHuggingFace Papers

Self-Evolving Coding Agents: Agents That Keep Improving After Deployment

This paper surveys self-evolving coding agents — systems that continue improving after deployment by learning from real dev feedback like test failures and repair attempts. The key implication: future coding agents won't stay static; they'll compound experience just like human developers do.

📚
AI PapersHuggingFace Papers

Agent Workflow Resume Semantics Are Broken Across 5 Major Frameworks

Researchers machine-checked five major agent workflow frameworks and found their crash-resume behavior is inconsistent and violates their own stated contracts. Critical warning for production agent builders: your workflow after a crash may not behave the way you think it does.

🛠️
AI ToolsLatent Space

Jeff, Sanjay, Oriol, Quoc Leave DeepMind — What's Happening at Google AI?

Following Jeff Dean, Sanjay Ghemawat, Oriol Vinyals, and Quoc Le have all departed DeepMind. Latent Space's deep-dive highlights a triple crisis at Google AI: compute politics, product strategy confusion, and a brain drain. Demis moves to Chair — the org shift has major industry implications.

Subscribe for daily AI updates + free playbook

📘 Subscribe Free