ChatGPT and Google Gemini have rolled out some seriously powerful new features, and the biggest shift is simple: these tools are becoming much better at actually getting work done. From Gemini watermark controls and web research to ChatGPT computer history, browser control, skills, and Google Drive editing, AI is moving far beyond basic chat.
If you use AI for content creation, research, coding, operations, email, or repetitive admin work, these updates can make your workflow way faster. The opportunity is not just generating better text. It is building systems that can understand context, work across tools, and help you complete multi-step tasks without living in 15 browser tabs.
Google Gemini Can Control Media Watermarks
One of the most surprising Gemini updates is a setting for generated media watermarks. When you create an image, video, or music asset in Gemini, generated content may include a watermark by default. Depending on the artwork, it may be subtle enough that it is difficult to spot visually.
For example, a simple prompt such as “Create a photo of a sloth” can produce an excellent image, but the downloaded file may include a small watermark. Gemini’s settings now include a Media Watermark control that lets you switch this behaviour off for future generated media.
How to Change the Gemini Watermark Setting
- Open Gemini and go to its settings.
- Find the Media Watermark option.
- Turn the setting off if you do not want watermarks included in newly generated images, videos, or music.
- Generate a new asset after changing the setting.
This matters if you create assets for social media, branding, or client work. A visible or machine-readable marker can affect how an asset is handled by platforms and detection systems. It is worth checking the settings before building a whole content pipeline around Gemini generated media.
That said, always make sure you understand the applicable platform rules and usage terms for the media you create. A setting change does not remove the need to use AI content responsibly.
Suggested image: A Gemini settings screen showing the Media Watermark toggle.
Suggested alt text: “Google Gemini media watermark setting for AI generated images videos and music.”
Gemini Spark Makes Web Research Much More Practical
Gemini Spark is where things get even more interesting. Instead of relying solely on general model knowledge, Spark can perform live web research and, when needed, use a Chrome extension to work through browser-based tasks.
Think about a request like this: find a Ferrari SF90 for less than $400,000. That is not a question with one static answer. Inventory changes, listings disappear, dealer prices vary, and sponsored placements can distort search results.
Spark can search across relevant sources, identify market pricing, surface listings, and make it easier to check the actual pages behind the results. In the Ferrari example, approved pre-owned listings were higher, around the mid-$400,000 range, while some marketplace listings appeared closer to the target price. That difference is exactly why live research is useful.
Where Gemini Spark Can Save Time
- Comparing current prices across marketplaces
- Finding products that meet a specific budget or requirement
- Researching competitors, offers, and industry trends
- Gathering sources before writing a report or content brief
- Completing browser-based research that would otherwise require multiple searches
The real advantage is speed. A task that normally means opening CarGurus, Cars.com, manufacturer pages, specialty marketplaces, and search results can be narrowed down much faster. You still need to validate the final details, especially for high-value decisions, but the research phase becomes dramatically more efficient.
Why Gemini 3.7 Flash Is Better for Agentic Tasks
Gemini 3.7 Flash is positioned as a much stronger model for agentic work. That means work where the AI has to take a goal, interpret a set of instructions, break the work into steps, and execute those steps efficiently.
There is a big difference between asking an AI to answer one question and asking it to complete a workflow. A workflow may require it to research information, decide what matters, use tools, organize findings, create a draft, and report back. The model needs to handle context over a longer chain of tasks without falling apart halfway through.
The update highlights stronger performance in areas such as production code quality, long-horizon software engineering, and enterprise workflow automation. In the workflow automation comparison discussed in the release material, Gemini 3.7 Flash outperformed prior Gemini versions and several competing models.
That does not mean you should hand every task to an AI agent blindly. It does mean that Gemini 3.7 Flash is a strong default when you need a model to follow a process, use connected tools, and move from task one to task four quickly.
Use Gemini 3.7 Flash for Work Like This
- Multi-step web research
- Software and coding tasks with several dependent actions
- Operations workflows that follow rules and decision paths
- Automations that require multiple tools or connected services
- Tasks where speed matters but context still needs to be preserved
ChatGPT Computer History Can Remember What You Were Doing
ChatGPT desktop has introduced a feature called Computer History, and this is one of the biggest changes in the entire update. When enabled, ChatGPT can record and take notes about activity on your computer, creating a searchable history of what you were working on.
For example, if you spend time preparing content, researching a topic, changing settings, or completing a process, ChatGPT can create notes from that activity. Those notes can be stored as local Markdown files, making it possible to revisit what happened later.
This opens up some genuinely useful possibilities. You could use it to preserve context across tasks, document a workflow as you perform it, remember what research you did earlier, or help an AI understand the background behind a request without explaining everything from scratch again.
Privacy Controls Matter Here
Obviously, a feature that records computer activity needs careful permissions. The good news is that you can control it.
- Exclude specific apps from computer history.
- Exclude specific websites from computer history.
- Pause recording whenever you are doing sensitive work.
- Resume recording when you are ready to capture useful context again.
The smart approach is not to turn everything on and forget about it. Use Computer History intentionally. Keep sensitive apps and websites excluded, pause the feature when needed, and treat permission settings as part of your normal workflow setup.
Build ChatGPT Automations With the Right MCPs, Plugins and Prompts
ChatGPT can automate a huge number of tasks, but there is one catch: it needs the right setup. That means choosing useful integrations, connecting the correct MCP servers, and giving the AI instructions that are specific enough to be reliable.
This is where LLMHelper AI can be useful. Its automation builder is designed to help map your role, the tasks you want automated, and the tools you already use into a practical AI workflow.
For a content creator, that might include:
- Writing video scripts
- Finding new content ideas
- Researching potential sponsors
- Responding to sponsor emails
- Organizing materials in Google Drive
You can specify tools such as VidIQ, Gmail, and Google Drive, then generate a recommended automation setup. The platform can identify connectors, suggest MCP servers, provide prompts, and even offer repair prompts if an automation does not work correctly the first time.
That repair prompt piece is underrated. Automations break. Permissions change, tools update, data formats shift, and prompts sometimes need refinement. Having a structured way to troubleshoot the workflow is far better than staring at a failed output and wondering what happened.
The goal is not to automate everything just because you can. Start with repetitive work that takes time but does not require your best creative judgment. Even a few well-designed automations can free up hours each week.
The ChatGPT Chrome Extension Brings AI Into Your Existing Workflow
The ChatGPT Chrome extension makes ChatGPT much more useful because it can work with the context already on your screen. Rather than copying an article into a new chat, explaining what it is about, and then asking for help, you can use the current page as the source context.
For example, you could open an article and ask ChatGPT to write a 45 to 60 second short-form script about the topic on screen. If ChatGPT has access to your established writing style or a saved script skill, it can use that context to produce something much closer to your usual format.
This is a massive quality-of-life upgrade for creators, marketers, researchers, and operators. It reduces the friction between finding information and turning that information into something useful.
Key ChatGPT Extension Features
- Screen context: Use the active page or visible information as part of a request.
- Memory and skills: Apply saved preferences and established formats to new tasks.
- File access: Pull in files when needed for a larger workflow.
- Plan mode: Give ChatGPT a goal and have it build an implementation plan.
- Goal mode: Set a target and allow the system to continue pursuing it.
- Record a skill: Capture your screen, voice, and actions so a repeatable process can be turned into an AI skill.
Record a skill is especially wild. If you repeatedly complete a process the same way, such as gathering data, organizing a file, or formatting an output, recording that workflow can help transform it into a reusable capability.
Just be realistic with goals. Asking an AI to build a full $10,000-per-month business from scratch is not a serious automation strategy. It can burn through credits and create a lot of messy work. Use narrow, specific, measurable goals instead.
Choose the Right Approval Level for ChatGPT Computer Use
When ChatGPT can use tools, access files, or interact with the internet, approvals become important. The available settings generally range from strict approvals to broader access.
- Ask for approval: Requires frequent confirmations, which provides more control but can make the workflow painfully slow.
- Approve for me: Allows routine actions while asking for confirmation on potentially unsafe actions.
- Full access: Gives ChatGPT broad access to the internet and files on your computer.
For most people, Approve for me is the practical middle ground. It keeps the workflow moving while still placing a checkpoint around actions that may be risky. Full access should only be used when you fully understand the scope of what the AI can do and have taken privacy seriously.
Run ChatGPT Locally or in the Cloud
ChatGPT can also run in different environments. You can run it on your computer, which allows it to work with local files when permission is granted. Or you can run it in the cloud, where it cannot access local files but can continue operating around the clock.
Local operation is useful when the work depends on documents, folders, or files stored on your machine. Cloud operation is useful when you need an ongoing task that you can manage from another device where you are signed in to ChatGPT.
This creates a more flexible setup:
- Use local access for file-based work and computer-specific tasks.
- Use cloud execution for long-running workflows.
- Use permissions deliberately based on the type of task.
- Keep sensitive data and critical actions behind appropriate approval settings.
Voice Hotkeys, Dictation and Screen Context Make ChatGPT Feel Always Available
ChatGPT desktop also includes expanded voice controls. You can set a voice chat hotkey, allowing you to open voice chat from anywhere on the desktop. There are dictation options for holding a key to dictate or toggling dictation on and off with a hotkey.
You can also build a custom vocabulary or journal for words and phrases that should be recognized correctly. That is particularly useful for names, product terms, technical language, or words that speech recognition often gets wrong.
Screen context adds another level. When you refer to something currently open, ChatGPT can inspect the foreground app and better understand what you mean. Combined with voice, this can make quick requests feel much more natural.
Edit Google Docs and Sheets Without Leaving ChatGPT
One of the best productivity upgrades is the ability to work with Google Drive directly inside ChatGPT Work. You can reference Google Drive, create or edit documents, and work with Sheets without constantly switching back and forth between tabs.
The document can sit alongside ChatGPT so you can ask for changes, generate content, reference information, and apply updates in one place. For anyone who scripts content, manages spreadsheets, organizes projects, or works from documents all day, that is a much cleaner experience.
Instead of juggling email, Google Drive, a script document, and separate research tabs, you can centralize more of the work in one environment. That is exactly where AI becomes useful: not as a novelty, but as a tool that removes tedious friction.
Use These AI Features to Save Time, Not Create More Chaos
The biggest takeaway from these ChatGPT and Google Gemini updates is that AI is becoming more operational. Gemini can research current information and act more effectively in multi-step workflows. ChatGPT can understand your computer context, use browser information, record skills, help manage documents, and automate connected tools.
But the best results come from starting small. Pick one recurring task. Define what a good result looks like. Give the AI clear tools, permissions, and instructions. Test it. Fix it. Then expand.
That is how you turn AI from something interesting into something that genuinely saves time every week.
For more AI tool breakdowns and workflow ideas, explore more resources from Rob The AI Guy and check out the full feature walkthrough. If one of these updates changes your workflow, share the article with someone who is still doing that task manually.
Frequently Asked Questions
Can Google Gemini create media without watermarks?
Gemini includes a Media Watermark setting that can be turned off for newly generated images, videos, and music. Check the Gemini settings before generating the asset.
What is Gemini Spark used for?
Gemini Spark is designed for web research and agentic tasks. It can search for current information, compare results, and use browser capabilities when needed.
What does ChatGPT Computer History do?
ChatGPT Computer History can record activity and create notes about what you were doing on your computer. You can pause it and exclude specific apps or websites for privacy.
Which ChatGPT approval setting is most practical?
Approve for me is a practical middle option because it allows routine actions while requesting confirmation for actions that may be unsafe.
Can ChatGPT edit Google Docs and Google Sheets?
ChatGPT Work can connect with Google Drive so you can create or edit documents and spreadsheets while working inside ChatGPT.
