Sunday Rundown #146: AI Design & Owl Twin
Live & Learn #6: Conversational AI image editing.
Happy Sunday, friends!
Welcome back to the weekly AI news roundup.
In case you missed it, here’s this week’s Thursday deep dive:
If you’re consistently missing out on my emails, remember to check your “Promotions” tab and mark whytryai@substack.com as a “Safe Sender.”
👩💻 AI releases
Adobe expanded its Creative Agent across Firefly and Creative Cloud apps, letting creators describe an outcome and have the agent handle the busywork.
Anthropic news:
Claude Design was upgraded so that it stays on-brand, syncs with Claude Code, and can export results to Adobe, Canva, PDF, and PowerPoint.
Claude Code now supports Artifacts that turn your session work into live, self-updating web pages that you can share with your team.
ByteDance launched Seedance 2.0 Mini, a faster, lower-quality version of its flagship AI video model that also costs half as much per generation.
Character.AI launched a Creator Bundle with new tools to help you build characters faster, track performance, and get discovered via leaderboards.
HappyOyster launched HappyOyster 1.0, an open-ended model for creating and exploring real-time interactive video worlds with prompts. (Try for free.)
Meta added AI Mode to Facebook search, along with new photo and video editing presets that let you swap looks, outfits, and styles.
Microsoft officially released Copilot Cowork, an AI teammate that can send emails, draft documents, manage your calendar, and browse the web.
Midjourney rolled out Draft Mode for V8.1, which lets you quickly generate 24 low-resolution image options before rendering your favorite in full resolution.
OpenAI news:
ChatGPT now gives free users more accurate and complete health answers using GPT-5.5 Instant.
Record & Replay lets you demonstrate your work steps to Codex and turn them into editable, reusable skills to share with others.
Scheduled tasks now have a dedicated pane in ChatGPT, making reminders and recurring prompts easier to create and manage.
Perplexity launched Brain for Computer, a memory system that learns from completed tasks to make the agent more precise and work better over time.
Sakana AI launched Marlin, an autonomous research agent that can think for up to eight hours straight to produce full reports and slide decks.
xAI released Grok Imagine Video 1.5, an image-to-video model with better motion consistency and faster generation speed.
Z.ai open-sourced GLM-5.2, an agentic model with a one-million-token context window, built for complex coding and long-horizon tasks. (Try it for free.)
📖 AI resources
“Americans and AI 2026” [REPORT]: Pew Research survey on how Americans use AI and their attitudes about it.
“Is It Agentic Enough?” [GUIDE]: Practical guide for benchmarking open AI models against your own tools and tasks from Hugging Face.
“LifeSciBench” [BENCHMARK]: A new evaluation suite from OpenAI to test how AI models handle real-world life-science research tasks.
🔀 AI random
SpaceX agreed to acquire Cursor-maker Anysphere for $60 billion in an all-stock deal, just as Cursor teased an upcoming launch of a new AI coding model.
🤦♂️ AI fail of the week
“Wow! Check out that beautiful view!”
-Hold up, lemme just dislocate my neck and feet real quick…
📹 Live & Learn #6: AI image editing
Live & Learn concept in a nutshell:
I pick 3-5 tools in the same category
I test them live and rate them on 3-5 specific dimensions
I end up with a leaderboard based on my observations and ratings
This week’s Live & Learn was all about conversational AI image editing:
In my next Live & Learn session, I’ll put AI models’ vision capabilities to the test.
Join me on Friday, June 26, at 2PM CET (8AM EST):
Also, please let me know what you’d like to get out of these live sessions by taking this quick chat-based survey.
See you there on Friday!