9 Hidden ChatGPT Settings You Need to Turn On Right Now
Most people are using ChatGPT at 20% of its actual power. I was one of them.
It was 2:17 AM.
I was staring at my laptop screen, third cup of coffee going cold beside me, watching ChatGPT spit out another answer that felt like it had been written by a tired intern copying from Wikipedia.
Same prompts. Same structure. Same boring outputs.
I’d been at this for weeks. Tried different prompt formulas. Watched YouTube tutorials. Read Reddit threads where everyone swore they’d “cracked the code.” Nothing worked. The answers kept coming back flat, generic, weirdly formal — like talking to someone who technically speaks your language but somehow never gets you.
I was close to giving up. Not on AI entirely, but on the idea that ChatGPT could actually be useful for the kind of nuanced, context-rich work I was trying to do.
And then, almost by accident, I clicked something I’d walked past a hundred times.
Settings.
Not the API settings. Not some hidden developer mode. Just the little gear icon sitting there in the sidebar, patiently waiting for me to finally pay attention.
What I found inside changed how I use ChatGPT completely.
Here’s what nobody tells you: most people are using ChatGPT with almost all its best features turned off, misconfigured, or completely ignored. The default setup is built for the average first-time user — cautious, stripped down, with no knowledge of who you are or what you actually need.
But underneath that surface? There’s a whole other version of the tool. Smarter. More personal. Almost eerily good.
I’m going to walk you through the 9 settings that unlocked it for me. In order. The way I found them.
Grab something to drink. This one’s worth reading to the end.
Setting #1: Custom Instructions — The One That Changes Everything First
I almost missed this one because the name sounded boring.
Custom Instructions. It sounds like a setting for adding your name to a form. But when I actually opened it, I realized I’d been doing something genuinely stupid for months.
Every single conversation I’d had with ChatGPT, I was starting from zero. No context. No background. No sense of who I was or what I was trying to accomplish. I was the stranger at the counter, ordering the same coffee every day, and the barista never once remembered my name.
Custom Instructions gives ChatGPT a permanent briefing about you.
There are two boxes. The first asks: “What would you like ChatGPT to know about you to provide better responses?” The second asks: “How would you like ChatGPT to respond?”
Most people leave them blank.
Here’s what I put in mine. First box: I told it I write content for tech-savvy but non-technical audiences. I mentioned I work across multiple projects. I specified that I prefer concise answers with examples, not theory. Second box: I asked for responses that skip unnecessary preamble, avoid hedging language, and treat me like someone who’s done their research.
The difference was immediate.
Before Custom Instructions, ChatGPT would open every response with “Great question!” and two paragraphs of context I didn’t need. After? It started mid-thought, like picking up a conversation with a colleague who already knew the backstory.
Why most people ignore it: It requires writing a few sentences about yourself, which feels like effort. And the feature isn’t prominent — it’s tucked inside the settings menu, not advertised anywhere on the main screen.
Why you shouldn’t: This single setting probably delivers more value than everything else on this list combined. Do it first.
Setting #2: Memory — Letting It Actually Learn You
Once I had Custom Instructions set up, I discovered the feature that should logically come next.
Memory.
What it does sounds simple: ChatGPT remembers things you tell it across conversations. But the implications of that are quietly enormous.
Before I turned this on, every conversation existed in isolation. I’d explain my project, explain my audience, explain my preferences — and the next day, I’d do it all again. It felt like a relationship with permanent amnesia.
With Memory enabled, ChatGPT starts building a profile over time. Not in a surveillance way — you can see exactly what it’s stored and delete anything you want. It remembers things like: “User prefers structured lists for complex topics” or “User is working on a content site targeting US professionals.”
I tested this a few weeks in. Started a fresh conversation, typed nothing but “help me write an intro for my next article” — and it responded with something that already reflected my tone, my audience, and my usual format.
I stared at that for a long moment.
This is what people mean when they talk about AI that actually adapts to you. It’s not magic. It’s Memory.
To find it: Settings → Personalization → Memory. Make sure it’s toggled on. Then check “Manage Memory” every few weeks to see what it’s learned and clean out anything outdated.
The caveat: It learns gradually. If you want to speed it up, occasionally say things like “Remember that I prefer X” and it will add it explicitly.
Setting #3: Personality and Tone Preference — Because ‘Formal’ Was Killing Me
Here’s the thing about ChatGPT’s default tone.
It’s professional. Relentlessly, exhaustingly professional. Every response has the cadence of a carefully reviewed email from a corporate communications team. Which is great if you’re writing press releases. Terrible if you’re trying to write something that actually sounds human.
Inside Settings, there’s a relatively new option to adjust response style and tone. On some interfaces it’s framed as a communication preference — you can push ChatGPT toward being more casual, more direct, more concise, or more conversational.
The moment I shifted this away from the default, my outputs transformed.
Same prompt. Completely different energy. The sentences got shorter. The vocabulary felt more natural. The responses stopped reading like they’d been optimized for legal review.
This matters more than people realize, especially for writers. When you’re using ChatGPT to help draft content — articles, emails, copy, social posts — the style of the AI’s output bleeds into your own work. If it’s stiff, your edits are stiff. If it’s loose and natural, you have something you can actually build with.
The hidden insight here: ChatGPT is a collaborator. And like any collaborator, you need to communicate how you like to work together. The tone preference setting is just that conversation — made permanent.
Setting #4: Data Controls — The Setting Everyone Should Check (But Almost Nobody Does)
I want to be upfront: this one isn’t about making ChatGPT smarter. It’s about making sure ChatGPT isn’t learning things you don’t want it to learn.
Under Settings → Data Controls, there’s an option called “Improve the model for everyone.”
By default, on some account types, your conversations may be used to help train future versions of the model. A lot of people don’t know this. Fewer still have consciously chosen whether they’re okay with it.
If you’re pasting in client work, confidential business information, personal health details, legal documents, or anything sensitive — you may want to turn this off.
I’m not saying it’s dangerous. OpenAI has privacy safeguards in place. But this is one of those settings where informed consent matters. You should be making a deliberate choice, not just living with the default.
To check it: Settings → Data Controls → toggle off “Improve the model for everyone” if you want more privacy.
The other thing you’ll find here: The option to export all your ChatGPT data. Useful if you’ve been using it for months and want to review, archive, or analyze your own conversation history.
Not glamorous. But important.
Setting #5: Chat History and Searchability — You’re Losing Work You Already Did
This one hit me harder than I expected.
I’d been having great conversations with ChatGPT. Genuinely useful sessions where we’d worked through something complex together, and by the end I had exactly what I needed. Then I’d close the tab, start a new chat a few days later, and have no idea how to find that previous conversation.
Get The AI Simplifier’s stories in your inbox
Join Medium for free to get updates from this writer.
I knew it was somewhere. But ChatGPT’s default search experience for your history is… not great.
The setting here isn’t a toggle so much as a habit you need to build around: making sure Chat History is enabled (it’s under Settings → Data Controls) and understanding how to use the search bar in the sidebar.
What I didn’t realize was that when history is on, every conversation is indexed and searchable by keyword. I’d been scrolling manually — like someone looking for a physical file in an un-alphabetized cabinet.
Once I started actually using the search field? I could find any conversation within seconds. Old research sessions. Draft frameworks. Prompt experiments that had worked well. They were all there — a personal archive of every useful thing I’d produced.
Why this matters for ChatGPT productivity: Your best prompts are a resource. Your most useful conversations are a library. Treat them that way.
If Chat History is turned off, none of this is saved. You’re starting from zero every single time, even the parts you shouldn’t have to repeat.
Setting #6: Temporary Chat — The Secret Sandbox Mode
There’s a feature that works in the opposite direction from everything I just described.
Sometimes you don’t want ChatGPT to remember the conversation. Sometimes you’re working through something speculative, sensitive, or half-formed. You want to think out loud without that thinking becoming part of your permanent profile.
That’s what Temporary Chat is for.
Enable it (it’s in the top-right area of the interface, not buried in settings, but worth knowing about) and your conversation won’t be saved to your history and won’t influence your Memory. When the chat ends, it’s gone.
I use this in a few specific situations: when I’m brainstorming ideas I’m not ready to commit to, when I’m testing a new prompt framework before I’m sure it works, or when I’m working with content that’s professionally sensitive.
The parallel here: Think of it like an incognito tab. Same browsing power, no trail left behind.
Where people get confused: Temporary Chat isn’t a privacy guarantee in the absolute sense — it just means ChatGPT won’t store or learn from it on your end. OpenAI’s own data handling policies still apply. But for day-to-day working purposes, it gives you a genuinely clean slate.
Use it strategically. Don’t make it your default — you lose the benefits of Memory and history. But know it’s there when you need it.
Setting #7: Voice Settings — The One That Actually Surprised Me
I resisted this one for a long time.
Voice features felt gimmicky to me. Like asking your calculator to also play music — technically possible, but not why I’m here.
Then one afternoon I was stuck on a piece of writing and too tired to type. Almost as an experiment, I switched to Voice mode and just… talked through the problem. Explained what I was working on. Described what wasn’t working about my current draft. Asked questions out loud the same way I’d ask a colleague.
The response came back as audio. And something about hearing the answer instead of reading it changed how I processed it.
I’ve since figured out when Voice actually works better than text for me: when I’m brainstorming (speaking is faster than typing when thoughts aren’t fully formed), when I’m reviewing something and want it read back (catches problems my eyes skip), and when I’m in motion — walking, cooking, or in between tasks.
Inside Voice Settings (Settings → Speech), you can choose between different voice options, adjust the response speed, and configure whether Voice mode activates automatically or only on demand.
Pick the voice that doesn’t make you want to turn the volume off after 30 seconds. There are several options. Some are noticeably better than others — take five minutes to try them all.
The surprising productivity use case: Dictate rough drafts verbally into ChatGPT. Then ask it to clean up the transcript into structured prose. It’s faster than you’d think, and the output is more natural than typing directly because you’re not second-guessing every word as you go.
Setting #8: Advanced Reasoning / Thinking Mode — The Feature That Changes What “Difficult” Means
This is the one I saved for near the end because it requires a bit of context to appreciate.
Standard ChatGPT gives you fast answers. It’s optimized for speed and fluency. Most of the time, that’s exactly what you want.
But there’s a class of problem where fast isn’t good enough. Complex analysis. Multi-step reasoning. Technical problems with a lot of moving parts. Creative challenges that need genuine strategic thinking rather than pattern-matched response.
For those, there’s a Thinking Mode — sometimes labeled o1, o3, or simply “Reasoning” depending on your plan and interface version.
When you switch to this mode, ChatGPT slows down. You can often see it actually thinking before it responds — working through the problem step by step rather than jumping to an answer.
The difference in output quality on hard problems is significant.
I tested it on a competitive analysis I needed for a project — a task where the wrong answer would actually cost me time and money. Standard mode gave me something plausible and surface-level. Reasoning mode gave me something I would have been embarrassed not to have figured out myself.
The tradeoff: It’s slower. And depending on your plan, you may have a limited number of Reasoning queries per day or month. Don’t use it for simple tasks. Save it for the questions that actually deserve more than a quick answer.
How to access it: Look for a model switcher at the top of your chat window, or a toggle that says “Reasoning” or shows the o1/o3 model. The location varies slightly by interface.
Setting #9: Connected Apps and Integrations — ChatGPT Is Not an Island
The last thing I found in my 2 AM settings exploration was the one that reframed how I thought about ChatGPT entirely.
It’s not just a chatbot. It’s an operating layer.
Under the integrations and plugin settings (now often called “Tools” or found inside GPT Builder options depending on your access level), ChatGPT can connect to external services: browsing the web, accessing files, running code, connecting to data sources, working with tools you already use.
This is where ChatGPT features expand from “text generator” into something more like an actual AI assistant.
When I enabled web browsing, ChatGPT stopped giving me answers from its training data alone and started pulling current information — live prices, recent articles, up-to-date documentation. When I used the file-reading capability, I could drop in a PDF and have a conversation with the contents.
The integration that changed my workflow most: The ability to upload a document, ask ChatGPT to analyze it, pull out key points, and then help me write something in response — all in a single conversation. No copy-pasting between tabs. No losing context mid-task.
Why most users don’t set this up: It requires a few minutes of exploration inside the Tools or Integrations panel, and the interface has changed a few times as OpenAI has updated it. But the current version is simpler than it used to be. Worth finding.
One important note: With more integrations enabled, you’re giving ChatGPT broader access to take actions and retrieve information. Take five minutes to review what’s turned on and make sure you’re comfortable with the permissions. The tool is useful. You should still understand what it’s doing.
The Part Nobody Tells You
Here’s what I’ve realized, sitting on the other side of all these settings.
ChatGPT out of the box is built to work for everyone. Which means it’s optimized for no one in particular.
Every setting I described above is ChatGPT’s way of asking: who actually are you, and what do you actually need? Most people never answer the question. They take the defaults, get generic results, and conclude that AI is overhyped.
It’s not overhyped. It’s just unconfigured.
The difference between the tool I was using at 2 AM that night and the tool I use now isn’t ChatGPT’s capability. It’s the 30 minutes I spent telling it who I am, how I think, and what I need from it.
That 30 minutes has paid back many times over.
Your Move
Open ChatGPT right now. Click the gear icon. Start with Custom Instructions.
Write two or three sentences about your work and how you like to receive information. Turn on Memory. Adjust the tone. Check your Data Controls.
You don’t have to do all nine settings in one sitting. But do the first two today. Notice what changes.
Then come back and do the rest.
The version of ChatGPT you’ve been using isn’t the only version. It’s just the one you haven’t configured yet.
If this helped you, consider sharing it with someone who’s been complaining that “ChatGPT isn’t that useful.” They’re probably running on defaults.
Follow The AI Simplifier for more practical, no-hype guides to getting real value from AI tools.
