← All Conversations
July 05, 2026

You are using ChatGPT completely wrong

ChatGPT productivity AI tools custom instructions thinking mode

Summary

Most people treat ChatGPT as a glorified autocorrect, asking it only to shorten emails or fix typos, and consequently reap only marginal gains. The author realized that this approach is akin to driving a Formula 1 car like a bicycle—missing the vehicle’s true power. The breakthrough came when he stopped requesting superficial edits and began using ChatGPT as a thinking partner: to plan, build, and solve problems that had stalled him for weeks. This shift unlocked dramatically higher productivity, allowing him to accomplish before lunch what once took a full day. The difference between novice and expert use boils down to a handful of concrete practices. First, leveraging ChatGPT’s custom instructions can fundamentally upgrade every interaction. By inserting the Metacognitive Reasoning Prompt—a framework derived from MIT research on recursive reasoning—the model is forced to decompose complex problems, assign confidence scores to sub‑solutions, verify logic and facts, combine answers with weighted confidence, and reflect on weaknesses before responding. For simple queries it defaults to a direct answer, but for anything intricate it delivers responses that are measurably sharper, more thorough, and over 110 % more accurate than standard prompting. Activating this prompt in the settings menu ensures that every subsequent conversation benefits from this higher‑order reasoning. Second, ChatGPT’s built‑in app store transforms the chat box into a hub for external services. Users can connect directly to platforms such as Spotify, Canva, Booking.com, Instacart, DoorDash, OpenTable, Expedia, Zillow, and dozens more without leaving the conversation. In practice, this means planning a trip—searching flights, reserving hotels, drafting itineraries, and recommending restaurants—all within a single chat, or designing graphics via Canva, ordering food through DoorDash, or checking real‑estate listings via Zillow. Most users never discover this integration, yet it eliminates context‑switching and streamlines workflows. Third, matching the model’s thinking mode to the task’s importance prevents wasteful over‑ or under‑use. ChatGPT offers Instant, Light, Standard, Extended, and Heavy thinking modes accessible at the bottom of the chat box. For quick facts or casual queries, Instant or Light suffices. For business plans, strategic decisions, complex writing, coding, or deep analysis, switching to Extended or Heavy engages step‑by‑step reasoning, self‑checking, and logical validation, yielding outputs of substantially higher quality. The author advises reserving the heavier modes for work that truly matters and reserving the lighter modes for trivialities. Applying these three practices—embedding a metacognitive reasoning prompt in custom instructions, activating relevant app‑store integrations, and deliberately selecting the appropriate thinking mode—shifts ChatGPT from a passive editing tool to an active cognitive partner. The result is not just incremental efficiency but a qualitative leap in output quality and speed, positioning users ahead of the majority who still treat the model as a simple autocomplete. By adopting these habits today, one can expect to be in a markedly different productivity category by tomorrow.

Hey

Okay, I have to be honest about something embarrassing.

For the first few months I used ChatGPT, I was basically using it as a fancy autocorrect.

Type a sentence. It fixes it. Done.

That was my entire workflow. I am not joking.

Someone would send me a long email and I would paste it in and say "make this shorter." That was my pro tip. That was my big AI unlock. I genuinely thought I was ahead of people because at least I was using it.

I was not ahead of anyone.

I was treating a Formula 1 car like a bicycle. Getting on it, pedaling slowly, and wondering why the people around me were moving so fast.

The shift happened when I stopped asking ChatGPT to do tiny edits and started asking it to help me think. To plan. To build. To solve problems I had been stuck on for weeks.

Now I get more done before lunch than I used to do in a full day. And the gap between how I use it and how most people use it comes down to a handful of specific things.

Here are 5 of them.

Most people reading this will know maybe one. If you apply all five today, you will be in a completely different category by tomorrow morning.

Best Nano Banana Use: 100+ Cases

100+ Nano Banana Use Cases is a ready-to-use prompt pack that unlocks the full creative power of Google's Nano Banana AI image model. Get over 100 tested prompts, options, and real-world examples t...

$9.99 usd: https://www.cooldeep.ai/products/best-nano-banana-use-cases

Tip 1: Put This Prompt in Your Custom Instructions

Most people have never touched their custom instructions. It is sitting there in ChatGPT settings, completely empty, doing nothing.

This is the single biggest upgrade you can make right now.

There is a prompt called the Metacognitive Reasoning Prompt. It came out of MIT research on recursive reasoning. What it does is force ChatGPT to slow down before answering you. Instead of firing off the first thing it comes up with, it breaks your problem into smaller pieces, rates its own confidence in each answer, checks its logic, and only gives you a response once it has verified its own work.

The result? Responses that are sharper, more thorough, and significantly more accurate. Research shows it outperforms standard prompting by over 110%.

Copy this and paste it into your ChatGPT custom instructions right now:

“_Adopt the role of a Meta-Cognitive Reasoning Expert. For every complex problem:_ _1. DECOMPOSE: Break into sub-problems_

_2. SOLVE: Address each with explicit confidence (0.0–1.0)_

_3. VERIFY: Check logic, facts, completeness, bias_

_4. COMBINE: Using weighted confidence_

_5. REFLECT: If confidence is below 0.8, identify weakness and retry_ _For simple questions, skip to direct answer._“ —

Go to ChatGPT settings, click "Customize ChatGPT," and paste it under the instructions box. Every single conversation from that point forward is a different experience.

This is the tip most people reading this will skip. Do not skip it.

Tip 2: ChatGPT Has a Full App Store and Almost Nobody Uses It

This one surprised me when I first found out about it.

ChatGPT has a built-in app store. You can connect it directly to Spotify, Canva, Booking.com, Instacart, DoorDash, OpenTable, Expedia, Zillow, and dozens more. All without leaving the chat window.

Think about what that means in practice.

You are planning a trip. You can have ChatGPT search flights on Expedia, find hotels on Booking.com, build a day-by-day itinerary, and suggest restaurants on OpenTable. In one conversation. Without switching tabs. Without opening five different apps.

You want to design something. Connect Canva. Describe what you want. It starts building.

You are ordering food. Connect DoorDash. Ask it what is near you. Done.

People think ChatGPT is just a chat box. It is not. It is a hub that connects to the tools you already use and runs them together for you.

Go to the ChatGPT app store, search your most-used apps, and connect them today. Most people never find this. You just did.

Tip 3: Switch On Thinking Mode for Anything That Actually Matters

Here is the mistake almost everyone makes.

They use the same ChatGPT setting for every task. Quick question about the weather. Deep strategy session for their business. Same mode. Same output quality.

That is like using a hammer for everything, whether you need a screwdriver or a drill.

ChatGPT has a Thinking mode. You will see it right at the bottom of the chat box. Click on it and you will find options: Light, Standard, Extended, Heavy.

When you are working on something that matters, a business plan, a decision you have been going back and forth on, a complex piece of writing, coding, analysis, switch it to Extended or Heavy.

What happens is ChatGPT actually reasons through the problem before it responds. It thinks in steps. It checks its own logic. The answer you get back is a completely different quality than what you get from a quick Instant response.

The difference is not small. It is massive.

Use Instant for quick things. Use Thinking mode for things that count. Once you start doing this, you will not understand why you ever used it without this on.

### 10 AI Stocks to Lead the Next Decade

AI isn’t a tech trend – it’s a full-blown, multi-trillion dollar race, and 10 companies are already pulling ahead.

These are the innovators driving real revenue, attracting institutional attention, and positioning for massive growth.

Get all 10 tickers in [The 10 Best AI Stocks to Own in 2026](http://marketbeat.com/newsletter/pdfoffer.aspx?offer=10BestAIStocks&RegistrationCode=Beehiiv-BUBDNZO6MS-10BestAIStocks&_bhiiv=opp_c8f740f3-db3e-4281-8973-015b7733319f_7d4f8797&bhcl_id=b2ec905c-4d0c-4acf-8bdf-e48c24e9398e_f672f2b6-1e60-4422-8448-ee13d3aa1099_22c35bc7-5e17-4448-baa5-49c4bf23f93e), free today.

[Download Now](http://marketbeat.com/newsletter/pdfoffer.aspx?offer=10BestAIStocks&RegistrationCode=Beehiiv-BUBDNZO6MS-10BestAIStocks&_bhiiv=opp_c8f740f3-db3e-4281-8973-015b7733319f_7d4f8797&bhcl_id=b2ec905c-4d0c-4acf-8bdf-e48c24e9398e_f672f2b6-1e60-4422-8448-ee13d3aa1099_22c35bc7-5e17-4448-baa5-49c4bf23f93e)

##### **Tip 4: Stop Starting New Chats. Use Projects Instead.**

This is one I added myself and it changed my whole workflow.

Most people treat ChatGPT like a search engine. Open a new tab, ask a question, close it, forget it. Every conversation starts from zero. ChatGPT has no idea who you are, what you are building, or what you talked about last time.

Projects fixes this completely.

Go to the left sidebar in ChatGPT. You will see a "Projects" section. Create one for every major thing you are working on. Your business. Your content. Your job search. Your side project.

Inside a project, ChatGPT remembers everything. Every conversation you have had. Every file you have uploaded. Every preference you have told it. You do not repeat yourself. You do not re-explain your situation. You just continue where you left off, like working with an assistant who never forgets.

I have a project for CoolDeep AI. It knows my tone, my newsletter structure, my goals, my audience. Every time I open it, it already knows the full context.

This is how you stop using ChatGPT occasionally and start using it as a real tool.

Set up one project today. Just one. See what happens.

Tip 5: Give It a Role Before You Ask It Anything

This is the simplest tip on this list and possibly the most powerful one.

Most people type their question and hit send. They get a generic answer. They wonder why ChatGPT does not feel that useful.

The fix is one sentence before your actual question.

Tell it who to be.

"Act as an experienced marketing strategist who has worked with early-stage startups."

"Act as a financial advisor helping someone with no investing background understand the stock market."

"Act as a brutally honest editor who does not sugarcoat feedback."

When you assign ChatGPT a role, everything changes. It adjusts its vocabulary. Its assumptions. Its depth. Its tone. It stops being a generic answering machine and starts behaving like a specialist you hired for exactly this problem.

Compare these two prompts:

_Without a role:_ "How do I grow my Instagram?"

_With a role:_ "Act as a social media growth expert who has grown accounts from 0 to 100,000 followers in competitive niches. My account is about AI and productivity. Give me a 30-day growth strategy with specific daily actions."

Same tool. Completely different answer.

Try it on your next prompt. You will never go back to asking plain questions again.

##### My Final Words…

When I was using ChatGPT to fix typos, I thought I was doing something. I was not.

The tool itself is free. The knowledge of how to actually use it is what separates the people winning with it from the people barely touching it.

Custom instructions. The app store. Thinking mode. Projects. Role prompting.

Five things. Most people use none of them.

You just learned all five.

Now open ChatGPT and try one of these today. Not tomorrow. Not after you finish reading everything else. Today.

The gap between you and the 99% closes the moment you actually apply this.

Stay curious, talk to you tomorrow.

### Most AI content is fascinating. None of it is useful.

You’ve read the AI breakdowns. Watched the million AI explainers. Nodded along to the hot takes.

And then opened a blank doc and had no idea what to actually do.

[The Shift ](https://magic.beehiiv.com/v1/2a941ba2-3b7f-49da-ab31-7c7f148b52e4?email=bob@reisner.info&redirect_to=https%3A%2F%2Ftheshiftai.beehiiv.com%2Fc%2Fwelcometotheshift&utm_source=beehiiv&utm_campaign=BUBDNZO6MS&redirect_delay=3&_bhiiv=opp_83970eb9-2012-444f-9008-0991a3b04317_d5df7e06&bhcl_id=e85a8645-f4cd-44ce-ab8c-abca78060bfe_f672f2b6-1e60-4422-8448-ee13d3aa1099_22c35bc7-5e17-4448-baa5-49c4bf23f93e) is a newsletter built to help you in moments like this. Every tool covered works in the real world, and every prompt in the 1000+ library solves something you’ll actually hit.

[3,000+ vetted tools. 1000+ tested prompts.](https://magic.beehiiv.com/v1/2a941ba2-3b7f-49da-ab31-7c7f148b52e4?email=bob@reisner.info&redirect_to=https%3A%2F%2Ftheshiftai.beehiiv.com%2Fc%2Fwelcometotheshift&utm_source=beehiiv&utm_campaign=BUBDNZO6MS&redirect_delay=3&_bhiiv=opp_83970eb9-2012-444f-9008-0991a3b04317_d5df7e06&bhcl_id=e85a8645-f4cd-44ce-ab8c-abca78060bfe_f672f2b6-1e60-4422-8448-ee13d3aa1099_22c35bc7-5e17-4448-baa5-49c4bf23f93e) Daily newsletter.

And right now, **3 subscribers win a free 1-year Claude Pro subscription. **One click to enter.

_[Subscribe to The Shift](https://magic.beehiiv.com/v1/2a941ba2-3b7f-49da-ab31-7c7f148b52e4?email=bob@reisner.info&redirect_to=https%3A%2F%2Ftheshiftai.beehiiv.com%2Fc%2Fwelcometotheshift&utm_source=beehiiv&utm_campaign=BUBDNZO6MS&redirect_delay=3&_bhiiv=opp_83970eb9-2012-444f-9008-0991a3b04317_d5df7e06&bhcl_id=e85a8645-f4cd-44ce-ab8c-abca78060bfe_f672f2b6-1e60-4422-8448-ee13d3aa1099_22c35bc7-5e17-4448-baa5-49c4bf23f93e)_

*Partner