Big News of the Week
ChatGPT just dropped v. 5.2
So what actually changes with ChatGPT 5.2?It just got released yesterday, so I haven’t time to play with it. According to my research, Here’s the short version of what you’ll supposedly notice
1. Different “modes” for different jobs
Instead of one model trying to do everything, 5.2 splits things up:
- Instant: fast answers, quick drafts, everyday stuff
- Thinking: slower but better when you need logic, strategy, or multi-step reasoning
- Pro: the heavy hitter for complex, professional work
Translation: you get more control over speed vs depth vs accuracy instead of hoping one setting magically does it all.
2. It can handle way more context
This is what I’m most excited about. 5.2 can keep track of extremely long conversations and documents, like full books, contracts, multi-month projects, giant SOPs.
This matters if you’ve ever said, “Wait, you already forgot what I told you.”
3. Fewer confident wrong answers
Early reporting shows about a 30 percent drop in hallucinations compared to previous versions.
Apparently t’s noticeably better at staying grounded on complex topics instead of confidently making things up.
4. Stronger for real work, not just chatting
This is where they are saying 5.2 really shines. It’s better at:
- Spreadsheets and models
- Technical and structured documents
- Presentations and user-experience thinking
- Coding and debugging
- Multi-step planning with tools
5. Smarter with visuals
It’s better at understanding things like charts, diagrams, and screenshots. You can also generate visual outputs more directly, which is useful for explaining ideas, workflows, and layouts.
6. Tools work more reliably
When you connect ChatGPT to tools, APIs, or large data sources, 5.2 is better at:
- Pulling the right information
- Staying consistent
- Not breaking halfway through a task
That’s a big deal if you’re using it for anything beyond simple prompts.
Claude Code Is Coming to Slack
Claude will soon debug code, generate snippets, propose fixes, and explain errors right inside Slack threads. Instead of switching tools, your team can stay in the conversation while Claude helps refine the code in real time. It turns Slack into a lightweight development space and makes technical collaboration smoother.
Google Introduces Disco, an Experimental Browser That Builds AI Widgets for You
Disco is a new Google Labs experiment designed to "take the web for a fresh spin" by exploring next-generation AI features for browsing. Its debut tool, GenTabs, turns your open tabs into custom, interactive apps tailored to your specific goals—whether you are planning a travel itinerary, designing a meal plan, or researching a complex topic. It looks super cool, based on what I see here.
You can join the waitlist here to get early access.
...and while we are talking about cool new Google tools, check this out:
While I was fiddling around with Disco, I saw this thing called Mixboard. I’ts an experimental AI-powered "concepting board" from Google Labs designed to help you visualize and refine creative ideas. It functions as an infinite canvas where you can combine uploaded assets with AI-generated visuals to brainstorm everything from interior design to marketing campaigns.
Key Features:
- Visual Iteration: Use the new "Nano Banana" editing model to modify images using natural language (e.g., "turn this into apple cider") or generate new variations with a single click.
- Smart Canvas: The interface allows you to organize loose ideas and images spatially, similar to a digital mood board.
- Instant Presentations: When your idea is ready, a "Transform" feature can automatically convert your messy board into a structured slide presentation.
You can try the experiment here: Mixboard.
Google Brings Gemini to Chrome on iPhone and iPad
Google is rolling out Gemini inside Chrome on iPhone and iPad, giving you quick AI assistance directly in your mobile browser. You can summarize pages, rewrite text, draft messages, and get explanations without switching apps.
Google Tests AI Article Overviews in Google News
Some Google News stories now show short AI-generated summaries as soon as you open them. The idea is to help you grasp key points quickly when you’re scanning several articles. It’s efficient, but I’m afraid things like this are going to make us really dumb.
Google Shares Security Details for Chrome’s Upcoming Agentic Tools
Google explained how Chrome’s new agentic features will stay secure as they begin completing tasks for you, like navigating sites, filling forms, or organizing research. These tools will run inside isolated sandboxes so they can help without accessing your private data. You’ll be able to decide exactly what the agent can see and do.
Instacart and OpenAI Bring Grocery Ordering Directly Into ChatGPT
You can now browse recipes, pick meals, swap ingredients, compare options, and send everything to Instacart — all inside ChatGPT. You stay in one conversation from meal planning to checkout, making grocery shopping much more streamlined. As someone who relies on Instacart constantly, I have to admit I’m a bit skeptical about how smoothly this will run at first. I already hit plenty of glitches with Instacart as it is (there was that time I wanted 12 apples, but accidentally purchased 12 BAGS OF APPLES, so I’m curious whether this new integration will simplify things or add a few new twists along the way.
Don’t miss out: Did you miss our last newsletter? We featured Google’s new Workspace Studio, a set of AI agents that work directly inside Gmail, Docs, Sheets, Slides, and more. They can summarize, draft, rewrite, organize, generate slides, and handle routine tasks with zero setup. The entire system is designed to understand the context in your files so you can get work done right where you already are. Check it here! |