Daily AI Tooling Roundup – July 04, 2026

June 2, 2026 Codex is becoming a productivity tool for everyone OpenAI today released a new report ⁠ (opens in a new window) , The Next Era of Knowledge Work , showing how Codex is no longer just a coding tool.. Increasingly, it’s helping people across professions automate…

Beyond LoRA: Can you beat the most popular fine-tuning technique?. +65 When you plan to fine-tune a model in a parameter-efficient way, think beyond LoRA If you want to fine-tune an open model on your own data, you are probably interested in so-called parameter-efficient…

Artificial Intelligence Fine-tune Amazon Nova models for accurate email data extraction The authors would also like to thank Karan Bhandarkar, Sue Cha, Yash Shah and Nieves Garcia for their contributions in making this initiative possible.. If you process millions of email…

June 1, 2026 Our views on AI policy and political advocacy AI is going to be one of the most powerful technologies people have ever built.. The decisions being made now about how it is governed and deployed will matter for a long time.. We believe that the future of AI should…

From the Hugging Face Hub to robot hardware with Strands Agents and LeRobot +10 A walkthrough of the LeRobot integration in Strands Robots – one agent loop, from a Hub dataset to a physical robot, with sim-to-real datasets in the same on-disk format and policies you swap with…

Artificial Intelligence How Amazon Bedrock catches AI-generated phishing Social engineering through phishing remains one of the most common tactics for launching cyberattacks.. AI-generated phishing email messages now pose a new challenge for security teams managing email…

June 30, 2026 Inside Genebench-Pro A closer look at the benchmark, its questions, and supporting materials.. Case studies These 10 case studies showcase representative questions from GeneBench-Pro.. Each case study includes the original prompt, datasets, and supporting…

GLM-5.2: Built for Long-Horizon Tasks +113 We’re introducing GLM-5.2, our latest flagship model for long-horizon tasks.. It marks a substantial leap in long-horizon task capability over its predecessor GLM-5.1 and, for the first time, delivers that capability on a solid…

Artificial Intelligence Structured memory filtering with metadata in AgentCore Memory Let’s say your customer support agent asks for “billing issues”, and gets back technical support tickets, sales conversations with receipt issues, and billing disputes all mixed.. This is…

June 9, 2026 How engineers at Nextdoor use Codex to build without limits On Nextdoor’s core platform team, Codex investigates issues and puts product engineers in the driver’s seat.. A product like Nextdoor, which serves over 110 million users across 11 countries, puts many…

Adding MCP Tools to Reachy Mini +18 The Reachy Mini conversation app can now use tools hosted in public Hugging Face Spaces, called over MCP.. You can give your robot a new ability, like checking the weather or searching the web, by adding a Space from the Hub instead of…

Artificial Intelligence How Inscribe uses Amazon Bedrock to stop document fraud in seconds This post is co-written with Conor Burke, CTO and Co-Founder at Inscribe Fraud now appears in 1 of every 16 documents, and AI-generated forgeries grew 5x from April to December 2025 (…