Let’s get this out of the way: AI is not just for tech bros who drink too much coffee and don’t get enough sunlight. It’s also for the rest of us: the thinkers, the scribblers, the overwhelmed and the overachievers. If you’ve ever thought: ‘I’m not technical enough for this’ … you’re exactly who this list is for.
It’s for you, me, your colleague who hates writing emails, your client who wants everything yesterday, and the part of you that is so tired of doing research that takes hours. And no, you don’t need to be a developer. You just need a brain, some spare time and a bit of curiosity to boost your own productivity. This is my absolute top 10, because admit it: you guys love lists.
1. ChatGPT, my OG that changed everything
Before Claude became poetic, Perplexity became your favorite nerdy librarian and Grok turned anime into a very weird UX-feature, there was ChatGPT: my tab of truth and chaos. We have been BFF’s since the start. But… you have to know how to use it wisely: otherwise it will always give you no fluff and that terrible: ‘You are not this, you are that’... 😩
Speed is not the problem. Garbage is. LLM’s overconfidently bullshit
The real trick? Super prompting. I know that all these aspiring ‘AI-millionairs’-stragetists say a single ChatGPT prompt will save your workflow and your business. Well, let’s stop pretending: anyone who’s really using large language models knows they hallucinate. Speed is not the problem. Garbage is. LLM’s overconfidently bullshit. They mirror bias. They miss nuance. Your first output is a mirror made of probability.
Loops to spot the liars
Super prompting isn’t about shouting into the void. You don’t just ask ChatGPT to ‘write a blog post about X’. You build logic into the prompt, ask it to think step-by-step, critique its own assumptions, test different mental models. You teach the AI to also meta-think, because that’s how YOU think smarter, not only faster.
And that's exactly your secret doorway: the point of using AI isn’t to find the answer, that’s what Google does for us. It’s to start a conversation you couldn’t have had without it.
When I use AI, I don’t trust the first response, I test it. I feed the same prompt into multiple models: ChatGPT, Gemini, Claude 3.5 Sonnet, Perplexity, Mistral. I let them disagree with each other. I watch where they converge. I track which claims show up across systems and which ones are outliers. I ask them to critique each other. I interrogate their citations. I check if the sources even exist. Every loop adds a layer of reliability.
Does your AI help you think better?
So when I hear people say: “AI helps me work faster” … I think: sure. But can it help you think better? Because that’s what matters. Not how quickly you produce, but how deeply validated your AI-workflow is. So what does that look like? Start with a problem you want to solve. Compare outputs from different models with specific meta-questions. Now you do what most people never do: you don’t choose.
You ask:
Where do they contradict?
Where do they converge?
Who adds useful citations?
Who fabricates or overstates?
Who makes implicit assumptions I didn’t ask for?
You can even feed their outputs back into each other:
“Claude, critique the reasoning steps in GPT’s output.”
“ChatGPT, identify where Claude overgeneralizes or is vague.”
“Perplexity, give me citations that directly confirm or contradict this claim.”
Then extract the structured essence into a JSON schema. That’s basically a structured format - like an intelligent checklist - that forces clarity. It tells the model exactly what goes where, and leaves no room for vagueness. Once you’ve gathered the best fragments - verified, refined, reflected - you now abstract it into JSON-prompts. Why? It’s the language that LLM’s understand most.
It forces you to commit to structure
You can plug it into downstream systems (UI, workflows, data pipelines)
You eliminate ambiguity
Rating 5/5
Price: Free
Notes: Because I work with ChatGPT 24/7, I use the Plus-subscription at 21,34 euro. For all of the other LLM’s like Grok, Gemini and Claude I use the free version.
2. NotebookLM, my way too clever assistant who knows it all
NotebookLM is my second set of eyes. Sometimes I can be a bit of a nutty professor: I feel like a hurricane with WiFi. I open 84 tabs in two minutes, chase five research trails at once, and forget what my original question was. That’s why I really like this app: NotebookLM calmly builds a mental map of what I’m trying to understand, even when I haven’t quite figured that out yet.
When I think of myself as that hurricane, NotebookLM is the quiet librarian who just walks around after me, picking up all the loose papers, color-coding the chaos, and whispering: “This bit here? That’s the insight you’re looking for!”
I want data, not the tweetable version
It doesn’t pretend that the author’s name matters more than the insight of certain research, and that makes it very interesting. I use it to upload everything: dense policy reports, whitepapers, long-read articles, academic PDFs, podcasts, interviews, YouTube videos. I ask: what exactly did the expert say about X Y Z? How does that connect to previous research we just disgusted? What was their actual conclusion: not just the tweetable version?
A key feature of NotebookLM is that its responses are grounded in the source material. It provides citations to the exact quotes or passages in the document where it found the information. This allows you to fact-check its answer and see exactly where the data came from.
NotebookLM filters the noise for me. It shows me the threads worth pulling
What were the actual conclusions? What was the methodology? What stood out? What did this researcher say that others don’t? NotebookLM filters the noise for me. It shows me the threads worth pulling. And while it’s busy loading a pile of data, I make coffee. Or do some yoga stretches. Or open Perplexity for a quick heads-up on all the AI-rumors. And the best part? The loop does its work.
And if I get to something out of NotebookLM that I don’t understand, I throw it into Explainpaper. That AI-assistant translates academic jargon into something I have the brain capacity for. I don’t let Google SEO decide what’s important by what ranks highest. I let AI help me figure out what’s actually useful for me on this topic.
Rating 5/5
Price: Free
3. Why I use Perplexity (and not Google) when things get serious
Have I mentioned that I like to research?🤓 I really like research, and therefore Perplexity is also part of my inner tech stack-circle. When I’m digging into a topic - whether it's a new AI paper, a health tech trend, or some obscure behavioral study - I don’t want a list of blue links and a migraine.
I want answers. Fast. I’ve thrown 10-part follow-up questions at Perplexity and it keeps the thread. It remembers the context. It evolves with me as I research. It doesn’t forget what we talked about two prompts ago. It doesn’t serve me SEO-optimized clickbait.
Google’s business model is powered by advertising. This means that search results pages are often cluttered with sponsored content and ads, which can sometimes push organic, relevant results further down the page.
Ten blue links. Click, scan, back button, click again. Repeat. And honestly, my brain is too tired for that
Perplexity doesn’t make me hop through ad-stuffed pages. It gives me a clean, sourced summary, right in the chat. I see exactly where it got its information from. I can click through, double-check, explore. Google, on the other hand, is still playing the 2010-game. Ten blue links. Click, scan, back button, click again. Repeat. And honestly, my brain is too tired for that. It’s a system built for ads, not for my brain that wants fast immediate gratification and answers.
Rating 4/5 (Because let’s be real: I want one app to rule them all. Right now I’m juggling ChatGPT, Claude, NotebookLM, Perplexity and my brain all at once. Merge them, please).
Price: Free
4. I got a little Goblin who helps me with my endless to-do-list
Imagine this: it’s Sunday night. My head’s full of half-thoughts, project ideas and imaginary sticky notes with “!!!” written on them. The same 84 tabs are open and one of them is playing music I can’t find. It’s not that I’m disorganized: my brain is too good at remembering everything all at once.
At 19.00 PM I always go to MagicToDo, the star of the Goblin Tools. It’s a to-do list-app, but nowadays it feels like my smart little AI-midget who follows me around everywhere. I don’t need to know how to structure my week, I need to dump a tangled sentence like: “Prep keynote, follow up, fix the prompt doc, pitches, creative projects, admin hellscapes???’’ … into something.
Make it extra spicy
And this ‘goblin’ calmly, sweetly chops it into bite-sized pieces. The spicier I rate the task (read: there are actual peppers?! you can click on when it becomes more stressful), the more finely it slices the steps. It’s amazing. And the magic? It does what my own brain sometimes refuses to: it creates order.
Doable order: it fixes my executive functioning. By the time I finish my tea, I’ve got a Monday plan, a Thursday email-draft, and a Friday wrap-up reminder, all sorted and emotionally ‘de-spiced’.
It even estimates the time I need for it. So when people ask how I stay organized, sane, and sharp while juggling ten AI models, strategy decks, and Substack essays all at once… I point to the gremlin in my browser tab.
Rating 4/5 (The interface is very minimal: some people will love that, I crave a more deep UX-experience).
Price: Free
Notes: I paid 1,99 euro for the mobile because it works better than the free browser. We all love the comfort of doing everything on our phone, right?
5. Compose AI writes most of my boring emails
I write fast, but I also write a lot. Articles, research blurbs, pitch decks, emails, prompts libraries, outlines, concept notes, scripts, call summaries. My brain’s always ahead of my fingers. That’s where Compose AI comes in, especially for those things I DO NOT want to write actually.
This is a writing assistant for your browser that autocompletes as you type, can generate text on demand, rephrase or shorten and draft email replies. It autocompletes as you type. It drafts responses.
Let the bots have it
Yes, it processes your browser data, like almost every AI productivity tool in this day and age. To be honest, I don’t mind that my Amazon purchase-return-question is written by an AI while I make coffee. So I might as well let my bot handle mine too.
If you write 1,000+ words a day like I do - across multiple tones, audiences, and formats - this extension is a timesaver. Its free plan includes 1500 AI‑generated words per month, which is great for testing, but a bit tight for heavy users.
Rating 4/5 (It still has access to your data, which might be a dealbreaker for some).
Price: Free
Notes: You get over a thousand words in the free subscription package. That’s a bit meh for a writer.
6. Elicit, the genie I use for health and biohacking
If you know me even a little, you know this: health is my obsession and biohacking is my love language. I had to battle long-covid for 3 years so experimenting with new ways to boost my bodies capabilities is always on my to-do-list.
The only thing is: PubMed is a nightmare of 83-tab rabbit holes. Health research papers are usually very difficult to understand and it takes ages to find the exact report you need to verify your biohacking-experiments with actual data.
No half baked Instagram-health-claims
Elicit is my personal health genie. I ask questions like: “How does early sunlight affect cortisol production in women?” or “What are the long-term effects of glycine on REM latency?” … and within seconds, it pulls peer-reviewed studies, summarizes the key findings, and links me directly to the source.
I’m not a doctor. I’m a nerd with questions
I ask Elicit for studies, cross-reference them in NotebookLM, and validate them with my own body data, such as lab results, my Apple Watch and Oura Ring. The best part is I don’t need to sift through 400 pages of PubMed. I can upload up to four papers at once and have a natural-language conversation with them. I ask: “Did they test women and men separately?” or “Was this double-blind?”
I’m not a doctor. I’m a nerd with questions. And Elicit gives me answers in a world full of half-baked Instagram-health claims and chatbots that make things up. This helps you because it derives its data from Semantic Scholar’s: this index contains over 200 million publications about science, social studies, engineering, medicine, and more.
Rating 5/5
Price: Free
7. Scholarcy keeps me sane in a world of AI chaos
Let’s be honest: the AI space is a firehose. Models drop faster than we can remember their names, newsletters recycle the half-understood hype, and every LinkedIn-guru seems to have ‘cracked’ the next big thing. If you DM them AI, they help you solve all of your IT-problems like a magical AI-fairy.
It’s overwhelming to keep up with the latest rumors, features, models and more. There’s a lot of bulk, reheated takes, junk and old trends that are recycled with some tech savvy sauce from Chat. And I don’t need to know everything about AI, just what actually matters to me.
This helps in a world where Google Search still thinks page 12 holds the answer
So how do I filter out the noise? I use Scholarcy as my personal research filter. Every Monday, right before I am off to the gym, I drop the week’s flood of newsletters, whitepapers, and transcripts into Scholarcy. It summarizes what I need: emerging trends, all things prompt engineering breakthroughs, natural language processing, ethical news and controversies, new AGI claims, model releases, and strategic implications.
Scholarcy handles bulk documents. It can process up to 64 papers or newsletter-style docs in one go. I use the Spotlight-feature to focus on what jumps out as curious or useful, not what the author decided to highlight. This helps in a world where Google Search still thinks page 12 holds the answer.
Rating 5/5
Price: Free (1 summary per day, so be prepared to junk-dump correctly)
8. Scraping the internet with Manus is the best research in history
Did I mention I love research?😂 With AI you can make your research 100 times faster if you use the concept of scraping. Instead of manually combing through dozens of sites, feeds, or news tabs one by one, scraping automates the gathering of information at scale and in real time.
That means you don’t just collect more data, you collect the right data faster. Once the raw flow is centralized, you can layer filters, ranking, and summarization with an LLM, turning thousands of scattered fragments into a coherent map of what’s happening.
I eat data for breakfast
Where old-school research gives you a glimpse of a few sources, scraping equips you with the whole picture, making your insights sharper, your trend-spotting quicker, and your strategic decisions up to a hundred times more effective because you’re working with breadth and depth simultaneously.
There are many tools out there - like Thunderbit, Browse AI and Kadoa - that can help you scrape Google News for example, but I like to work in LLM’s with my super scraping prompt.
I feed it my JSON-scraping-superprompt and it dives into the internet like a disciplined detective
The most advanced outputs I get are from Manus, developed by the Singaporean startup Butterfly Effect Pte. Ltd; it’s giving Chat Agent avant la lettre. Manus is an autonomous AI agent. I feed it my JSON-scraping-superprompt and it dives into the internet like a disciplined detective. Ten minutes later, I get a PDF full of high-signal insights: clean, sourced, and ready to feed into ChatGPT or Claude for deeper analysis.
Manus operates without without constant babysitting. It functions as a system of specialized sub-agents. A central ‘executor’ agent coordinates these sub-agents (e.g., a researcher, a content creator, and a verifier) to tackle different parts of a complex problem simultaneously, making it highly efficient.
Rating 5/5
Price: Free
9. Yapping into a microphone is a fulltime job
You wouldn’t say so, but I don’t always feel like writing. In fact, most of my ideas don’t show up when I’m sitting in front of a blank Google Doc. They arrive when I’m walking, vacuuming, doing groceries, or in the shower.
And it’s mostly chaotic, nonlinear, half-formed, but yet creative … so I don’t want to forget them. That’s where my Audiopen becomes essential. I open the app, hit record, and just speak. Rant, spiral. I let the noise out.
Do I feel a bit odd talking to my phone in aisle 7? Sure. Do I care? No, because it saves me at least two hours a day
Audiopen transforms my yaps into structured, cohesive notes that sound like me, but are more of an actual idea instead of an ongoing stream of words. It’s like having a voice assistant who knows my brain. It captures me and it organizes my thoughts and rearranges the chaos.
I can read it … and actually recall what I have said. Do I feel a bit odd talking to my phone in aisle 7 in the supermarket? Sure. Do I care? No, because it saves me at least two hours a day, and I never forget my best ideas.
Rating 4/5 (You cannot yap unlimited in the free subscription).
Price: Free
10. Building apps without being an app builder
I’m not a developer. I can’t code. I still get nervous when I see a Excel-spreadsheet with formulas. But I do have a brain wired for systems, habits, cycles, and checklists and nowadays, that’s enough. Rocket.new lets you build apps, landing pages and websites without writing a single line of code. Just with language. And that’s my superpower.
Strong writing is the real cheat code
My sentences enter the machine and get translated into vectors, neat rows of numbers. Behind the screen it’s pure nerd math. The cleaner your prompt, the more you’re basically feeding it perfect data. So yeah, strong writing is the real cheat code. And now you can put that writing into a web application that makes apps with it. The only limit is your imagination; and okay, your token budget.
I use it to build the web apps I need... like a supplements-tracker, a Qur’an study-buddy or a 12-month self-growth tracker that helps me integrate my goals, reflections, habits, all in one place. I use different LLM’s for the best prompt architecture that lies at the core of my app. I then create it as a web application and add that site to my iPhone homescreen.
Off course you still need to know what you are doing, and the first try always needs some tweaking … but it’s great for making ideas come to life, for your nerdy hobbies, or if you want to professionally pitch your demos.
Rating 5/5
Price: Free (you get up to 1 million tokens, good for testing, but not enough for final execution).
Notes: I paid 209 euro for the yearly plan, because… well, I build a lot of apps.
Well, this was it and hopefully it was helpful! Good luck on deep diving into the beautiful world of AI…
Happy prompting!
Sandra
Very good Suggestion 👏 👌