Sunday Rundown #112: Moodboards & Stabby Cat
Sunday Bonus #72: "AI Video Prompter" that turns ideas into finished prompts.
Happy Sunday, friends!
Welcome back to the weekly look at generative AI that covers the following:
Sunday Rundown (free): this week’s AI news + a fun AI fail.
Sunday Bonus (paid): an exclusive segment for my paid subscribers.
In case you missed it, here’s this week’s Thursday deep dive:
Let’s get to it.
🗞️ AI news
Here are this week’s AI developments.
👩💻 AI releases
New stuff you can try right now:
Adobe news:
Firefly Boards rolled out worldwide, with new editing and ideation tools and additional third-party video models from Runway and Moonvalley.
Photoshop’s Generative Fill feature now supports third-party models like Nano Banana (Gemini 2.5 Flash Image) and FLUX.1 Kontext.
Alibaba news:
Qwen3-Max is a massive trillion-parameter language model for coding and agentic tasks, with a “Thinking” variant in the works.
Qwen3‑Omni is an end-to-end multilingual model that can natively process text, image, audio, and video and respond in real time.
Anthropic brought its latest Claude models—Sonnet 4 and Opus 4.1—to Microsoft 365 Copilot enterprise for use with custom agents.
Genspark launched a photo-editing app called Photo Genius that lets you request edits by simply speaking to it. (Download for iOS or Android.)
Google news:
AI Plus plan rolled out to 40 more countries, giving access to Veo 3 Fast, built-in Gemini in Gmail, Docs, and Sheets, and higher NotebookLM limits.
Gemini 2.5 Flash and Gemini 2.5 Flash‑Lite both got a bump in quality, speed, and cost-efficiency.
Gemini for Google TV lets you chat with Gemini to discover new movies, get summaries, or ask questions.
Google Photos on Android got conversational editing, so you can make tweaks to pictures by just asking. (US only for now)
Mixboard is a brainstorming canvas that lets you create and edit idea boards using natural language. (Try here. US only, so might need a VPN.)
Search Live lets you talk to Gemini while sharing your camera feed to get troubleshooting tips, identify objects, and more. (US only for now)
Hailuo AI launched Hailuo AI Agent, an infinite canvas that understands your request and selects which image, video, or audio model to use. (Try it here.)
Higgsfield AI released WAN 2.5 with native sound effects and dialogue.
Kling AI launched Kling 2.5 Turbo, a faster and cheaper video model with accurate prompt following, smoother motion, and better style consistency.
Meta news:
Facebook Dating now has a dating chat assistant and a Meet Cute feature that finds your ideal matches. (Rolling out in the US and Canada.)
Vibes is an AI video feed where users can watch, create, and remix short AI clips in the Meta app.
OpenAI news:
ChatGPT Pulse provides tailored daily insights based on your chat history and connected apps. (For Pro subscribers only so far.)
Shared Projects in ChatGPT improve how teams collaborate with a shared memory, instructions, connected tools, and new admin controls.
Perplexity news:
Search API now lets developers tap into its answer engine for fast, structured results.
Email Assistant can draft replies, sort your inbox, schedule meetings, and more. (Available to Max users.)
Suno dropped v5, featuring immersive audio, more lifelike vocals, and better creative control. (For Pro and Premier users.)
WhatsApp added message translations so users can auto-translate chats across languages without leaving the app.
📖 AI resources
Helpful AI tools and stuff that teaches you about AI:
“GDPVAL: Measuring model performance on real-world tasks” [RESEARCH] - new benchmark from OpenAI to evaluate real-world performance.
“SEAL Showdown” [REFERENCE] - see how real people rate different models in this new benchmark by Scale AI.
🤦♂️ AI fail of the week
“Dude, chill. I’m pretty sure it’s already dead.”
Send me your AI fail for a chance to be featured in an upcoming Sunday Rundown.
💰 Sunday Bonus #72: Turn your vision into a full video prompt with AI Video Prompter
I’ve already made a bunch of prompt-assisting Claude Artifacts for you, including the “Smart Prompt Maker” for LLMs and the “Midjourney Prompt Builder” for images.
But what about AI video models?
That was a rhetorical question; you can stop screaming answers at your screen.
Today’s AI Video Prompter uses my home-baked “5S Framework” (see “Complete Beginner’s Guide to AI Video”) to help you go from a vague idea to a copy-paste prompt for any video model.
The Artifact works like this:
You describe your video idea.
AI Video Prompter asks you tailored follow-up questions to flesh out the details.
It fills out the 5S Framework and lets you edit it as you wish.
You get three prompt variations to test in your favorite video model.
Note: Using the AI Video Prompter requires a free Claude account.
Take it for a spin here: