Howdy, it’s Barsee.

Happy Monday, AI family, and welcome to another AI Valley edition.

Today’s climb through the Valley reveals:

• OpenAI now lets you adjust ChatGPT’s personality and tone

• Disney builds a walking Olaf robot that uses AI to prevent overheating

• Alibaba’s Qwen launches new image model with layered editing

• Meta is developing new AI image and video model code-named ‘Mango'

• Plus trending AI tools, posts, and resources

Let’s dive into the Valley of AI…

ADOBE

Adobe brings industry-leading apps directly into ChatGPT

Adobe is bringing Photoshop, Express and Acrobat directly into ChatGPT for FREE. Now, more than 800 million uses can edit photos, create designs and transform PDFs in the most intuitive way possible: using their words inside the chat.

For Photoshop: simply upload the photo into ChatGPT and start typing prompts for the desired result. Remove objects, replace backgrounds, and apply creative effects to photos simply by typing what you want. Customize effects with a simple slider for the perfect blend of easy use and image control.

For Acrobat: Upload and edit PDFs, extract text and summaries, convert formats, merge or split files, compress documents, and add watermarks or page modifications.

For Express: Search Adobe Express’ vast template library to find the best one for any moment, replace images and text and animate elements, to generate content on the fly.

Open up ChatGPT and try it out for yourself today!

_{*This is sponsored}

THROUGH THE VALLEY

1/ OpenAI now lets you adjust ChatGPT’s personality and tone

OpenAI is rolling out new controls that let you customize exactly how ChatGPT sounds. You can now tweak settings for warmth, enthusiasm, and emoji usage, or set preferences for formatting (like requesting headers or lists). This feature addresses feedback that the AI often sounds "fake friendly" or overly emotional. These settings don't change the AI's intelligence, just its delivery style. It is a major shift toward making ChatGPT a better tool for work, allowing professionals to get direct, serious answers without the "fluff." OpenAI is also reportedly looking to raise $100 billion, aiming for a massive $830 billion valuation.

2/ Disney builds a walking Olaf robot that uses AI to prevent overheating

Disney Research has engineered a fully autonomous robot of Olaf from Frozen, solving difficult physics challenges along the way. Because Olaf has a huge head, tiny body, and snowball feet, he is naturally unstable. To make him walk, the team built a robot with 25 moving joints managed by a smart AI system. The breakthrough here is an AI that monitors the motor temperatures in real-time: if the robot gets too hot, the AI adjusts its movements to reduce effort and cool down. This technology could be incredibly useful beyond theme parks, helping prosthetics and drones function better in tight, hot spaces.

3/ Alibaba’s Qwen launches new image model with layered editing

Alibaba has introduced "Qwen-Image-Layered," a model that changes how AI handles images. Instead of generating a flat picture, it builds images in separate, transparent layers (similar to professional design software). This means you can move, resize, recolor, or delete specific objects without distorting the background or the rest of the image. This approach allows for much cleaner edits, such as swapping people or fixing text, making it far more flexible and reliable than traditional pixel-based AI tools.

4/ Meta is developing new AI image and video model code-named ‘Mango'

Meta is working on a new generation of AI models scheduled to launch in the first half of 2026. The project includes "Mango" (focused on images and video) and "Avocado" (focused on text and coding). Led by Meta’s Superintelligence Lab, this initiative aims to drastically improve the AI's visual reasoning and coding skills. This is a high-stakes effort for Meta as it tries to catch up with competitors like OpenAI and Google. After recent leadership changes, the company is under pressure to prove it can deliver a breakout AI product.

TRENDING TOOLS

• Cto.bench > A ground-truth code agent benchmark built from real user work, measuring how Al agents perform on actual tasks rather than artificial or hypothetical problems

• Opal > A tool for building AI-powered mini apps is now directly available in the Gemini web app

• Sim > Open-source visual workflow builder for building and deploying AI agent workflows

• Manus > Edit slides created with Nano Banana Pro

• CarEdge > It tells you what to pay and what to say, then negotiates for you if you don’t want to

THINK PIECES / BRAIN BOOST

• Reflections on AI at the end of 2025

• 2025 LLM year in review

• AI #147: Flash forward

• AI agents are starting to eat SaaS - Martin Alderson

• Lenny Rachitsky says everyone should be using Claude Code more

• Why do AI chatbots use ‘I’?

• If you want to understand where AI is headed, don’t watch the benchmarks. Watch the bottlenecks

• Sam Altman on OpenAI's plan to win, AI personalization, infrastructure math, and the inevitable IPO

• Sam Rose explains how LLMs work with a visual essay

THE VALLEY GEMS

What’s trending on social today:

THAT’S ALL FOR TODAY

Thank you for reading today’s edition. That’s all for today’s issue.

💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee

👍️ New reader? Subscribe here

Thanks for being here.

REACH 100K+ READERS

Acquire new customers and drive revenue by partnering with us

Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.

If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.