ChatGPT and Google Gemini have both released a bunch of genuinely useful upgrades, and some of these changes are a much bigger deal than they first appear. We are talking about new privacy controls, usage analytics, multi-inbox email actions, AI sticker packs, book-powered research notebooks, generative video models, better transcription, and personal AI avatars.
There is a lot happening in AI right now, but these are the updates worth knowing because they can change how you actually use these tools day to day. Whether you are using AI for a business, content creation, teaching, coding, email, research, or just having fun, there is something here that can make your workflow better.
ChatGPT Temporary Chats Now Have Two Modes
One of the most interesting ChatGPT updates is the new way temporary chats work. Temporary chats are basically ChatGPT’s version of incognito mode. You can have a conversation without treating it like a normal, ongoing chat history.
The big change is that temporary chats can now be either personalized or unpersonalized.
Personalized temporary chats
A personalized temporary chat can still reference the things ChatGPT knows from your account setup, including:
- Your saved memory
- Your connected plugins
- Your custom instructions
This is useful when you want a conversation to remain separate from your regular history but still want ChatGPT to understand your preferences, background, working style, and the context it has learned about you.
For example, if you ask, “Who am I?” in a personalized temporary chat, ChatGPT may be able to provide a more relevant answer because it can use the memory and instructions connected to your account.
Unpersonalized temporary chats
Unpersonalized temporary chats do the opposite. They ignore your memory, plugins, and custom instructions. If you ask who you are, ChatGPT should not have the information necessary to answer based on your personal profile.
This gives you a clean slate. It is a great option when you want to test a prompt without your usual context influencing the response, work through a generic scenario, or simply keep the interaction more separate from your normal AI setup.
The really cool part is that a temporary chat does not have to stay temporary forever. If you have a conversation that turns out to be useful, you can save it. Think of it as being able to take something worthwhile from an incognito-style session and commit it to your regular ChatGPT experience afterward.
Suggested image: A screenshot of the ChatGPT temporary chat selector showing personalized and unpersonalized options.
Suggested alt text: “ChatGPT temporary chat settings with personalized and unpersonalized modes.”
ChatGPT Usage Analytics Make Your AI Habits Visible
ChatGPT also has a new analytics area in Settings that provides a clearer look at how you use the platform. This is one of those features that might not sound flashy, but it can be incredibly useful.
The usage history can show activity across different time frames, including the last seven days and the last 30 days. You can also break activity down by areas such as:
- Product: Which ChatGPT products or areas you use most
- Model: Which AI models are part of your workflow
- Surface: Where your activity is happening
- Tools and skills: Which tools you rely on most often
This is valuable because it gives you a real picture of your AI habits. You may discover that you keep reaching for one feature while barely touching another feature that could help you. You may also notice that a workflow you used regularly has disappeared from your routine.
That kind of visibility opens the door to more useful guidance over time. If ChatGPT understands what you are using, it could potentially identify areas you have neglected or suggest tools that match how you work.
AI is becoming more personal, and analytics are one more signal that these platforms are moving toward more customized experiences rather than one-size-fits-all chatboxes.
Turn Images Into Custom Sticker Packs With ChatGPT
ChatGPT’s image tools now include the ability to create stickers and sticker packs. This one is just fun, but it also has some very practical use cases.
You can upload an image and ask ChatGPT to create a sticker pack based on it. In the example, an image of a Ferrari was used to generate a collection of car-themed stickers. The result was a set of individual graphics that could be used digitally, printed, or repurposed for branding.
This feature can be useful for a lot more than novelty stickers:
- Creating branded sticker concepts for a business
- Making classroom graphics and reward stickers
- Designing kid-friendly art from family photos or favorite objects
- Generating social media graphics with a consistent visual theme
- Creating design ideas for merchandise, packaging, or print materials
The core idea is simple: you can take almost any visual starting point and turn it into a more expressive collection of graphics. That is a pretty neat capability to have built directly into an AI image workflow.
ChatGPT Can Connect Multiple Gmail Accounts
If you manage more than one inbox, this next ChatGPT update is probably one of the most useful. The Gmail integration can now support multiple Gmail accounts from one ChatGPT setup.
For anyone juggling a personal inbox, a work inbox, client inboxes, or multiple business accounts, that is a major quality-of-life improvement. Instead of treating each account as a completely separate system, you can bring those accounts into one AI-assisted workflow.
The Gmail actions shown include:
- Batch reading emails
- Batch reading email threads
- Batch responding to emails
- Connecting additional Gmail accounts
That means ChatGPT can become much more useful for email organization and response workflows. You can handle several inboxes without needing to jump constantly between accounts. For busy professionals, business owners, and creators, this has the potential to make ChatGPT dramatically more valuable.
Of course, email is sensitive. Before connecting accounts or using batch actions, take time to understand the permissions you are granting and review anything before it gets sent.
NotebookLM Can Now Add Eligible Google Play Books
Over on the Google side, NotebookLM has a very cool new source option: eligible books from Google Play Books.
NotebookLM is already useful because it lets you build a knowledge base from the sources you add. Now, if a book is eligible, you can add it as a source through the Google Play Books experience and bring that material into a notebook.
This makes sense for anyone building a research system around the books they are reading and studying. You could use a book as part of a larger collection of sources related to a topic, project, interest, or business problem.
For example, recommendations can align with subjects you are already interested in, such as coding, stoicism, decision-making, business, intelligence, food, and artificial intelligence. Books such as The Power of Habit and Thinking in Bets show the type of material that could fit naturally into this kind of personalized research environment.
How the Google Play Books integration works
- Find an eligible title in Google Play Books.
- Look for Gemini NotebookLM through the tools option on the book detail page.
- Create a new notebook or open an existing notebook.
- Select the book as a source.
- Use the book within your NotebookLM knowledge base.
Not every book is eligible yet, but the availability is expected to expand. The important part is where this is heading: AI research tools are becoming more connected to the material people actually learn from.
Suggested image: A NotebookLM workspace with a book imported as a research source.
Suggested alt text: “Google NotebookLM notebook using an eligible Google Play Book as a source.”
Google AI Studio Gets New Video and Transcription Models
Google AI Studio has added several models that are especially interesting for builders, developers, and people creating AI-powered workflows.
Gemini Omni 1.1 Flash for generative video
The new Gemini Omni 1.1 Flash video model is described as a production-ready generative video model. It supports keyframe control and scene extension, while allowing text, image, and video references to guide the output.
The model can generate videos up to 30 seconds long and can upscale results to 4K. That gives creators a lot more room to create polished short-form content, explainers, visual concepts, and branded video ideas.
There is one major consideration: cost. The video output pricing shown was equivalent to 10 cents per second at 720p, with a 30-second output costing about $17.50. Higher-resolution output can cost more.
That means prompt quality matters. When every generation has a meaningful cost, you want to think through the scene, references, style, camera direction, and desired outcome before you hit generate. Getting it right the first time becomes much more important.
Gemini 3.5 Transcribe and live transcription
Google has also introduced Gemini 3.5 Transcribe models, including a version optimized for pre-recorded audio files and static inputs, plus a lower-latency live version for real-time streaming audio.
The pre-recorded transcription model is designed to produce detailed, structured text along with rich archival metadata. The live option focuses on transcribing incoming audio quickly as it happens.
This is a big deal because better transcription means better video understanding, better document processing, and better systems for working with audio at scale. A document processor, for example, can become much more capable when it has strong speech-to-text behind it.
Inside Google AI Studio, these models can be explored through prompting, but their real power comes through the API. Developers can build agents, apps, transcription systems, and workflows that make these capabilities part of a larger product.
For more AI updates and practical walkthroughs, explore Rob The AI Guy’s channel and the original ChatGPT and Gemini update walkthrough.
Create AI Videos With Your Own Gemini Avatar
The final Gemini update is probably the one that will get the most attention: AI avatars.
Inside Gemini Settings, there is an Avatar area in beta where you can create an avatar of yourself. Once the avatar is created, you can tag it in a prompt and ask Gemini to generate a video around a specific topic.
For example, you could tag your avatar and ask for a video of yourself explaining E equals MC squared. Gemini can generate a video with the avatar in multiple scenes, speaking through the subject and appearing in an environment such as a whiteboard presentation.
This is pretty insane when you consider how close these tools are getting to usable personalized video production. The results are not perfect yet. Facial details, especially things like teeth, can still look a little off. But there is also a straightforward way to improve outcomes: refine the avatar inputs. If your reference photos show more variety or better expressions, the avatar may produce better results.
Once a video is generated, available options include:
- Sharing the video
- Showing captions
- Downloading the result
- Adjusting volume
- Editing elements through prompts
You can even request visual changes after generation. Want the avatar’s shirt to be purple? Tell Gemini, and it can make that adjustment. With models such as Omni, custom voice options can also add another layer of personalization.
Gemini also includes a media watermark setting that can be turned on or off. As AI-generated media becomes more common, controls like that are worth paying attention to.
Suggested image: An AI-generated presentation video featuring a Gemini avatar in front of a whiteboard.
Suggested alt text: “Google Gemini AI avatar presenting an educational explanation in front of a whiteboard.”
The Bigger Picture: AI Is Becoming More Personal and More Useful
These ChatGPT and Google Gemini updates are not just random features. Together, they point to a bigger shift.
ChatGPT is becoming more aware of how you work, from personalized temporary chats and usage analytics to connected multi-account email workflows. Gemini is becoming more capable of turning information, books, audio, images, and identity into usable outputs through NotebookLM, AI Studio models, transcription, video generation, and avatars.
The tools are getting more personalized, more multimodal, and more integrated into the things people already do. That is where the real value is.
Try one feature that matches a real problem in your workflow. Use temporary chats when you need a clean testing environment. Review your analytics to understand your habits. Build a notebook around a book you are studying. Experiment with an avatar for a quick educational video. The people who get the most out of AI are not necessarily using every new feature. They are finding the few features that save time, improve output, or unlock something they could not do before.
Share this article with someone trying to keep up with the latest ChatGPT and Gemini features, and keep exploring the AI tools that can actually make a difference in your day.
Frequently Asked Questions
What is the difference between personalized and unpersonalized ChatGPT temporary chats?
Personalized temporary chats can use your saved memory, plugins, and custom instructions. Unpersonalized temporary chats ignore those details and provide a cleaner, more separate AI interaction.
Can ChatGPT connect to more than one Gmail account?
Yes. The Gmail integration can connect multiple accounts and supports actions such as batch reading emails, reading threads, and batch responding.
Can NotebookLM use books as sources?
NotebookLM can add eligible Google Play Books as sources. Availability does not apply to every title, but eligible books can be included in a NotebookLM knowledge base.
What can the Gemini Omni 1.1 Flash video model do?
Gemini Omni 1.1 Flash can generate videos using text, image, and video references. It supports keyframe control, scene extension, videos up to 30 seconds, and 4K upscaling.
How do Gemini AI avatars work?
Create an avatar from the Avatar section in Gemini Settings, then tag that avatar in a prompt to generate videos featuring your likeness. Generated videos can be shared, downloaded, captioned, and edited through further prompts.
