Newsletter Archive

Hermes Agent, an open-source agentic framework from Nous Research, hit 140,000 GitHub stars in three months and is now the most-used agent globally on OpenRouter. It features self-evolving skills, contained sub-agents, and runs locally on NVIDIA RTX/DGX hardware; Alibaba's new Qwen 3.6 models (27B/35B) outperform much larger predecessors when deployed with Hermes.

NVIDIA AI · 6 min read Repos

Anthropic launched Claude for Small Business, a package of pre-built connectors and workflows embedding Claude into tools small business owners already use—QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and Microsoft 365. The offering aims to close the AI adoption gap for businesses representing 44% of U.S. GDP but lagging enterprise AI adoption.

Anthropic · 14 min read Industry

The popular ATM-didn't-kill-teller-jobs story obscures a harder truth: mobile banking wiped out teller employment in the 2010s by making bank branches obsolete. The parable economists use to reassure people about AI disruption may be fundamentally misleading.

DZone AI · 1 min read Community

Microsoft has invested over $100 billion into its OpenAI partnership, cementing one of the largest corporate bets on generative AI infrastructure and product integration.

Bloomberg Tech · 1 min read Market

OpenAI details how it built a secure sandbox for Codex on Windows, isolating file access and network permissions to let code generation agents run safely without breaking out into the host system.

OpenAI · 1 min read Tools

Anthropic is raising at least $30 billion at a $900+ billion valuation in what could be its largest funding round ever, positioning the AI safety-focused startup among the world's most valuable private companies.

Bloomberg Tech · 1 min read Market

Fastino Labs open-sourced GLiGuard, a 300M-parameter safety moderation model that matches or beats the accuracy of guardrails 23–90x larger while running 16x faster and at 16.6x lower latency. Built on encoder architecture instead of decoder-only design, it handles prompt safety, jailbreak detection, harm classification, and refusal detection in a single pass — directly addressing the operational cost chokepoint of running guardrails on every LLM request in production.

MarkTechPost · 9 min read Tools

Notion launched a developer platform enabling teams to embed AI agents, custom code, and external data sources directly into their workspace, positioning the company as a hub for agent-driven productivity software.

TechCrunch · 1 min read Industry