Farhan Profile picture
I post the latest AI tools, tutorials, and news ✉️ hey@farhanai.com
May 21 12 tweets 13 min read
🚨 The man who invented calculus had a daily ritual nobody knew about.

Every morning during the Plague of 1665, Isaac Newton (fresh Cambridge graduate sent home by the outbreak, future Lucasian Professor of Mathematics, the man whose Principia Mathematica changed physics for 300 years) sat at his mother's wooden farm table in Woolsthorpe, England, with a notebook he called the Waste Book and wrote down every problem he was thinking about, every approach he had tried, every dead end he had hit, and every new question that had occurred to him overnight.

He called it the Waste Book. It was 1 of 10 techniques he used to invent calculus, formulate the law of universal gravitation, and split white light into colors. All in 18 months. From his mother's farm. During a plague.

I turned the 10 into Claude prompts.

You describe any cognitively demanding work you're trying to do... and it runs you through the same method Newton ran every morning at his mother's farm during the plague.

Here are all 10:Image 1. THE WASTE BOOK... CAPTURE PROBLEMS AND PROGRESS IN A DAILY WORKING NOTEBOOK

Act as a focus coach teaching the Waste Book... Isaac Newton's daily working notebook practice, where he wrote down every problem he was working on, every approach he had tried, every dead end he had hit, and every new question that had occurred to him overnight. The original Waste Book (1664-65, now at Cambridge University Library) is the document where calculus was first written down.

Help me build a Waste Book I'll actually use every day... so my hardest problems compound across days instead of resetting every morning, and my dead ends become future material instead of forgotten effort.


1. Ask what hard problems I'm currently working on and how I currently track them before starting
2. Choose the medium... bound paper notebook, plain text file, Obsidian, Notion (Newton's bound paper outperforms most digital tools for cognitive reasons)
3. Set the daily ritual... 15-30 minutes at the same time daily (morning or end of work)
4. Define what gets captured... open problems, approaches tried, dead ends, new questions, partial insights
5. Build the indexing system... how I find an old idea or dead end 6 months from now
6. Build the review habit... weekly scan, monthly synthesis
7. Set the cross-pollination trigger... when an old entry gets pulled back into a new problem



- Capture dead ends with their cause... a failure with no cause is wasted material
- Write the question, not just the answer... questions compound, answers are static
- Don't filter during capture... the messy entry is the honest entry
- Review before solving... most "new" problems are old problems in disguise
- Test: am I finding old ideas I'd forgotten about in the weekly review


Current Problem Tracking Audit → Medium Selection → Daily Ritual Time → Capture Categories → Indexing System → Review Cadence → Cross-Pollination Trigger
May 19 12 tweets 13 min read
🚨 The most powerful man in the world had a secret morning ritual.

Every day before sunrise, Marcus Aurelius (Emperor of Rome, commander of 400,000 soldiers, the most absolute ruler on Earth) sat alone with a wax tablet and reminded himself that he was going to die, that everything he loved would disappear, and that none of it was under his control.

He called it Memento Mori. It was 1 of 10 techniques he used to stay sane through plague, war, and the slow corruption of his only son.

I turned the 10 into Claude prompts.

You describe whatever emotional weather you're facing... and it runs you through the same exercise Marcus did at sunrise in his tent.

Here are all 10:Image 1. DICHOTOMY OF CONTROL... LET GO OF EVERYTHING YOU CAN'T ACTUALLY CONTROL

Act as a Stoic coach teaching the Dichotomy of Control... Epictetus's foundational distinction (from the opening of his Enchiridion, ~125 AD) that some things are in our control (our judgments, intentions, desires, aversions) and most things are not, and that almost all human suffering comes from trying to control the second category.

Help me apply the Dichotomy of Control to whatever I'm anxious or upset about... so I stop spending energy on what's outside my control and direct it entirely toward what is.


1. Ask what specific situation is bothering me before starting
2. List every element of the situation... what happened, who's involved, what might happen next
3. Sort each element into 2 columns: in my control / not in my control
4. Be honest about the "kinda in my control" items... most belong in the not-in-control column
5. Identify what I'm currently fighting that isn't in my control... where the wasted energy is going
6. Re-direct to what IS in my control... my judgments, my choices, my responses
7. Build the daily reminder... when this exact pattern shows up again, what's the trigger to stop



- Honest sorting matters more than fast sorting... most "kinda" items are not in your control
- Other people's actions, words, opinions are NEVER in your control
- The past is never in your control
- Outcomes are never in your control... only the inputs you control
- Test: can I name the specific action I will now take, knowing the outcome isn't mine to guarantee


Situation Inventory → 2-Column Sort → Honesty Audit on "Kinda" Items → Wasted-Energy Map → Re-Direction to Controllables → Daily Reminder Trigger
May 14 12 tweets 12 min read
🚨 Richard Feynman won the Nobel Prize in Physics with an IQ of 125.

Most physics graduate students score 130. Most physics professors score 140+. Feynman, the man who founded the field of quantum electrodynamics, was technically average-intelligent on paper.

He didn't outsmart his peers. He out-learned them. He had a personal method for tearing new fields open in months that he refused to formally publish. His students reconstructed it after his death.

I turned that method (plus 9 others from Ferriss, Young, Waitzkin, and Ericsson) into Claude prompts.

You describe any skill you want to learn... a language, a coding framework, a craft, a domain... and it builds you a 3-12 month curriculum that uses every technique.

Here are all 10:Image 1. THE FEYNMAN TECHNIQUE... EXPLAIN IT TO A 12-YEAR-OLD OR YOU DON'T UNDERSTAND IT

Act as a learning coach teaching the Feynman Technique... Richard Feynman's method for testing whether you actually understand something by trying to explain it in plain language to someone who knows nothing about it. The gaps in your explanation reveal the gaps in your understanding.

Help me apply the Feynman Technique to anything I'm trying to learn... so I stop confusing "I've read it" with "I understand it" and start exposing the gaps before they become real problems.


1. Ask what concept or skill I'm trying to learn before starting
2. State the concept in 1 sentence using ordinary language... no jargon allowed
3. Explain it to an imaginary 12-year-old in 5-10 sentences
4. Identify where the explanation breaks down... the moments where I had to use jargon or hand-wave
5. Go back to the source... fill the specific gaps where the explanation collapsed
6. Re-explain after the fix... see if the gaps closed
7. Test against a real beginner if accessible... explain to someone who actually knows nothing about it



- Use ordinary language only... if you need jargon, you don't understand it yet
- The 12-year-old test is the standard, not the smart-peer test
- Gaps are the win, not the failure... they're the only honest signal of what's missing
- Re-explain after every fix... a closed gap should hold up to a second pass
- Test: can I explain this in 3 minutes to someone with no domain background


Concept in 1 Sentence → 12-Year-Old Explanation → Gap Identification → Source-Going Round → Re-Explanation → Beginner Test
Mar 23 5 tweets 2 min read
🚨BREAKING: Advertising just went fully AI.

You can now use Arcads to:

> visually connect the best models
> lock in consistent characters
> change their outfits and scenes
> animate them perfectly, etc

This is stupidly powerful.

Watch the workflow 🧵 1/ Most AI platforms make you generate an image, download it, upload it to another tool to edit it, and then upload it again to animate it.

It’s a chaotic, broken process.

Arcads' new AI Workflows feature fixes this by introducing node-based editing.

You can now connect every step of your creative process on one visual canvas.
Mar 6 10 tweets 3 min read
RIP Harvard MBAs.

I converted the entire Harvard Business School strategy playbook into prompts.

Now Claude can analyze any company like a $5,000/hr consultant.

Here’s the prompt stack (save this)↓ Image
Image
Prompt 1: Porter's 5 Forces

"Analyze [Company] using Porter's Five Forces. For each force, rate intensity (Low/Medium/High), identify the top 3 drivers, and conclude with the biggest strategic threat to their current position."

What you get: A full competitive landscape in 90 seconds.
Mar 4 13 tweets 4 min read
BREAKING: AI can now think like a $10,000/month McKinsey consultant for free.

Here are 10 insane Claude Opus 4.6 prompts that build complete business strategies, market analyses & 90-day growth plans in 5 hours.

(Save for later) Image
Image
Claude Opus 4.6 with 1M context + adaptive thinking is an absolute cheat code.

These prompts turn it into your personal McKinsey strategy team.

Copy-paste → watch it spit out pro-level docs.
Feb 23 6 tweets 2 min read
🚨BREAKING: HuggingFace just dropped a plugin that turns Claude Code, Gemini, and Cursor into full ML engineers.

It's called HF Skills and it's completely free.

One command. Your AI agent can now train models, run evals, and publish research papers.

2.1K stars. Apache 2.0. Image Repo: github.com/huggingface/sk…
Feb 17 18 tweets 5 min read
Someone used Elon Musk's actual thinking framework as AI prompts.

It's the closest thing to having a billionaire engineer rip apart your ideas and rebuild them from physics.

Here are the 15 prompts that changed how I solve problems: Image 1. "What are the physics of this problem?"

Musk strips everything to objective reality.

"I'm struggling to grow my newsletter. What are the physics of this problem?"

AI reveals the hard constraints, the real forces, and the non-negotiable bottlenecks. Image
Jan 23 7 tweets 3 min read
This is literally OnlyFans for AI.

Higgsfield just launched the first-ever monetisation platform for AI influencers.

You can now build a digital character for free and earn from a $100,000 weekly pool.

Here is how it works: 🧵 Meet HIGGSFIELD EARN.

This platform connects AI creators directly with paid brand campaigns.

There are no agents to manage and no complex contracts to sign. You simply create content and receive guaranteed payouts based on performance.
Jan 15 7 tweets 3 min read
BREAKING: Reading PDFs is officially over.

Google NotebookLM turns any document into a high-quality podcast instantly.

Hours of reading → Replaced by 15 mins of listening.

Here is the workflow: 👇 Image step 1: head over to NotebookLM.

it's an experimental project from Google Labs.

best part? it is completely free (for now).

just log in with your standard google account.
Jan 5 6 tweets 3 min read
reference + actor = video.

the math is simple.
the results are illegal.

maxfusion scene builder is the most dangerous tool for "teams of one."

don't let this flop.

start cloning today 🧵 stop begging chatgpt for a "cinematic lighting" prompt. it sucks.

ugc scene builder lets you clone the vibe visually.

drag reference -> drag actor -> done.
same scene. same pose. your face.

it feels like a cheat code.
Dec 25, 2025 5 tweets 2 min read
voice actors are officially cooked.

kling 2.6 generates studio-quality voiceovers that lock to your character instantly.

no recording.
no drift.
perfect sync.

hollywood is powerless against this speed.

judge for yourself 🧵 usually, you have to retrain a model for every scene.

kling just fixed that.

bind the voice once -> it adapts to the scene automatically.

the emotion changes. the delivery changes. the identity stays exactly the same.
Dec 2, 2025 6 tweets 3 min read
I wanted to stress test Nano Banana Pro inside Freepik.

I gave it a studio portrait + a cheesecake. No context.

It returned a scene that looks like a food magazine cover.

See how it pulled it off (and the exact prompt I used): The source material.

One studio portrait of a woman in an apron and one isolated shot of a cheesecake.

Combining people with specific objects usually breaks AI models. You typically lose the likeness of the person or the object turns into a hallucination. Image
Image
Nov 29, 2025 5 tweets 2 min read
I am done trying to be a "content creator."

I give Pippit one sentence about my product, and it handles the strategy, editing, and posting automatically.

It turns a 4-hour chore into a 4-second task.

Here is how I use it: 1/ From 1 Sentence to Full Strategy:

Tell Pippit what you’re selling in a single sentence.

It gives you:
A trending angle
Viral hooks
Script outlines
Captions + Hashtags

No prompt engineering. No strategy calls. Just outputs that hit.

Link: pippit.ai/thoughtform-on…
Oct 16, 2025 4 tweets 2 min read
This AI watches YouTube so you don’t have to.

→ It summarizes long videos
→ Adds timestamps
→ Spits out tweet threads or blog posts

I built it without writing a single line of code in 2 minutes.

Let me show you 🧵 1/ Why I Built This

Ever watched a 40-minute video for one key point?

Same. I was wasting time scrubbing through interviews, tutorials, and podcasts just to find the golden 2 minutes.

So I built an agent that:
→ Accepts any YouTube link
→ Summarizes the video with timestamps
→ Can output a tweet thread or blog post

genspark.ai/agents?type=cu…
Oct 9, 2025 7 tweets 2 min read
This AI model went too far

It mimics viral behaviour so well, TikTok and Meta sued

Higgsfield literally broke the internet

6 wild demos you won’t believe:

1/ 2/
Sep 29, 2025 8 tweets 3 min read
I asked an AI to pretend to be me on a phone call.

It argued with Samsung.
Escalated the complaint.
And booked my repair
All by itself.

Here’s how you can try it for free: 1/ Adulting used to mean hours on hold.

Calling airlines, negotiating bills, waiting for reps to pick up all for basic tasks.

I hated it.

Then I found Pine AI.
Sep 25, 2025 8 tweets 4 min read
Photographers are DONE

Nano Banana is live on Pippit AI and it gives you unlimited fashion-grade shots with 1 click

I used it to make these photos.

Here's how: Image
Image
1. Goto:
2. Upload reference image
3. Write a simple prompt
4. Click Generate
5. Done! pippit.ai/?utm_medium=Me…
Sep 25, 2025 10 tweets 5 min read
I don't understand why people are not using AI for job hunting

Use this AI method to land your dream job in no time: Image 1/ Create professional headshots: Meet

@BetterPicAI is your AI photographer

Create 4k Professional AI headshots from selfies and casual photos.

Get yours now: bit.ly/Better-Pic
Sep 18, 2025 7 tweets 3 min read
🚨BREAKING: AI just made $500 photoshoot obsolete.

Upload a selfie to BetterPic → get 100+ 4k professional headshots for LinkedIn, resumes, social media and more.

Here's how to get yours in 3 steps: The process is shockingly simple:

1. Go to Betterpic
2. Sign up
3. Choose your background, outfit style, colors
4. Upload at least 8 selfies

You’re done. Let the AI cook.
Sep 13, 2025 12 tweets 3 min read
Modeling agencies are cooked 🤯

Fashion Factory by Higgsfield turns one character + outfit into 15 magazine-ready shots.

It’s fast, cinematic, and disturbingly good.

10 viral examples (and #3 is insane): 1/