Zapier explains why human review still matters as AI moves from assisting people to running automated, multi-app workflows. The guide argues that agents miss nuance, edge cases and teams' unwritten rules, so people need checkpoints to catch mistakes before they spread.
Zapier has published a reference list of every AI model currently available for automation on its platform, grouped by provider. The guide is meant to help users keep up with a fast pace of model releases and pick ones worth adding to their workflows.
A developer known as Sac says he built a French-language 3D browser game based on The Little Prince in 16 days, using Claude Opus 5.5, Tripo and Blender. Korben looks at what the code reveals, the tools involved such as MCP for Blender and WebGPU, and why Linux users are left out.
Cities and local partners across Europe, the Middle East and Africa are using NVIDIA computing to build physical AI agents and digital twins. The tools are meant to help with everyday decisions such as scheduling road repairs or saving energy.
Hermes Agent 0.22 opens new desktop installs on a guided setup chat running on a free trial model, so newcomers can start before getting an API key. The release also adds connectors to everyday apps, a one-click llama.cpp runtime for local models, and stable versioning.
OpenAI profiles Third Moment Research, an AI lab that builds agents able to work on tasks for days and improve with use. The lab relies on GPT-6 Astra, GPT-6.1 Sol, Codex and Daybreak for cybersecurity.