scripod.com

Ep 802: ChatGPT’s Task Comeback, Claude’s Design upgrade, Codex Copies your workflow and 7 other Fresh AI features you’ll Want to use Today

In this episode, Jordan Wilson covers a wave of new AI features that are immediately available, including the return of ChatGPT Tasks, a major update to Claude Design, and a new open-weights model that tops benchmarks. The discussion highlights practical tools for users across different plans and use cases.
ChatGPT Tasks replaces the sunsetting Pulse, allowing paid users to schedule proactive background tasks like reminders and daily briefings using connected apps. Claude Design's June update introduces WYSIWYG editing, design system imports, two-way Claude Code integration, and export options to PDF, PowerPoint, Canva, and more. Google Vids upgrades AI avatars with Veo 3.1 and Gemini 3.1 Flash, offering 53 presets and unlimited video duration for US users. OpenRouter Fusion synthesizes outputs from multiple models via API, aiding research and development. Claude Code Artifacts now let team and enterprise plans create shareable visual pages from sessions. GLM 5.2 from ZAI, an open-weights model under MIT license, tops the Artificial Analysis intelligence index, excelling in coding and reasoning with a 1 million token context window. OpenAI's Codex Record and Replay records workflows to create reusable AI skills, competing with RPA tools.
00:00
00:00
All features are immediately usable.
02:36
02:36
ChatGPT Tasks replaces Pulse for proactive background tasks
07:47
07:47
Proactive AI agents work on your behalf
08:24
08:24
Great in theory but often failing in practice
09:22
09:22
Improved round-trip from prototype to production.
12:13
12:13
Integrates with Claude Code for design-to-code workflows.
15:23
15:23
Claude Design's new WYSIWYG editor offers fine-grained control
18:21
18:21
Google Vids offers AI avatars with directed actions.
24:00
24:00
Claude Code Artifacts turn session work into shareable visual pages.
27:41
27:41
GLM 5.2 tops the intelligence index.
29:09
29:09
GLM 5.2 excels in front-end coding and reasoning.