cd/daily·home›daily
grep --since=today /news | head -20

Daily Digest

Today's curated AI & tech news. Refresh hourly; emailed at 09:00 UTC.

today 20email subscribe →
23:49
superml.dev
artificial-intelligence

Your RAG Prompt Can't Fix Bad Retrieval

A benchmark from Jiutian Research at China Mobile, "EnterpriseRAG" (arXiv 2608.11584), found that 13 tested models including Claude Opus 4.5, Claude Sonnet 4, Gemini 2.5 Pro, GPT-4.1, DeepSeek and sev…

23:49
developers.googleblog.com
robotics

Gemini 4 Argon for Robotics

Google's Gemini 4 Argon scored 87% on the Embodied Reasoning (ERQA) benchmark, which tests a model's ability to solve physical, spatial problems in 3D environments, outperforming every other frontier …

23:44
blog.boxlite.ai
ai-agents

BoxLite v0.10.0: Your agent needs more than a sandbox

BoxLite released version 0.10.0, adding managed volumes, public inbound networking, and per-organization resource ceilings to its agent sandbox platform after 246 commits since 0.9.7. The release expo…

23:24
danielmiessler.com
ai-products

I Write My Own Stage Directions Now

Kai, Daniel's AI assistant, moved the spoken line at the end of every answer from ElevenLabs to Fish Audio's Drama 3 model, cloning its existing voice from a 30-second sample at 1.25× speed. Drama 3 a…

23:21
tezgrid.com
ai-infrastructure

TezGrid: AI inference on idle computers

TezGrid launched an AI inference service that routes standard OpenAI-compatible chat requests to idle consumer computers, letting machine owners keep 99.8% of each job while TezGrid takes 0.2%. Develo…

23:13
rustgis.app
ai-agents

RustGIS

RustGIS launched as a free, open-source GIS written in pure Rust that runs natively on Windows, macOS and Linux and in a web browser, keeping data local unless a user sends it to an AI model of their …

23:10
trigger.dev
ai-agents

Things we wish we'd known before building an AI chat agent

Trigger.dev published seven lessons from building its open-source dashboard chat agent, including a two-stage startup that cut the time to a useful update in half by streaming an immediate model call …