Chris Laub Profile picture
Oct 18 12 tweets 3 min read Read on X
R.I.P Google Scholar.

I'm going to share the 10 Perplexity prompts that turn research from a chore into a superpower.

Copy & paste these into Perplexity right now: Image
1. Competitive Intelligence Deep Dive

"Analyze [company name]'s product strategy, recent feature releases, pricing changes, and customer sentiment from the last 6 months. Compare against top 3 competitors. Include any executive statements or strategy shifts."
2. Technical Paper Breakdown

"Explain this paper [paste arxiv link or title] like I'm a senior engineer. Focus on: novel contributions, implementation feasibility, benchmark comparisons, and whether claims hold up under scrutiny. Skip the background fluff."
3. Regulatory Impact Analysis

"What are the latest regulations affecting [industry] in [region]? Summarize key compliance requirements, deadlines, penalties for non-compliance, and how major companies are responding."
4. Funding & Investment Trends

"Show me all funding rounds in [sector] from the last quarter. Include amounts, lead investors, valuations when available, and emerging patterns. Which sub-sectors are getting the most capital?"
5. Technical Benchmark Synthesis

"Compare performance benchmarks for [technology/framework] across: latency, throughput, resource usage, cost efficiency. Use data from official docs, third-party tests, and production case studies."
6. Executive Briefing Generator

"Create an executive summary on [topic] covering: current state, key players, market size, growth trajectory, risks, and strategic implications. Keep it under 500 words. Focus on actionable insights."
7. Patent & IP Landscape

"Research recent patents filed for [technology] in the last 2 years. Identify key patent holders, common approaches, white space opportunities, and potential infringement risks for building in this space."
8. Academic Consensus Finder

"What's the current scientific consensus on [topic]? Include peer-reviewed meta-analyses, systematic reviews, contradictory findings, and where researchers disagree. Flag studies with methodological issues."
9. Market Entry Analysis

"Evaluate entering [market/geography] for [product type]. Cover: market size, growth rate, key competitors, regulatory barriers, distribution channels, pricing expectations, and cultural considerations."
10. Real-time Event Monitoring

"Track all major developments related to [company/technology/event] from the last 48 hours. Prioritize primary sources, official announcements, and verified reports over speculation."
The difference between ChatGPT and Perplexity?

ChatGPT guesses based on training data.

Perplexity searches the current web, cites sources, and updates in real-time.

For research, that's everything.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Chris Laub

Chris Laub Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @ChrisLaubAI

Oct 10
Google just did the unthinkable.

They built a voice search model that doesn’t understand words it understands intent.

It’s called Speech-to-Retrieval (S2R), and it might mark the death of speech-to-text forever.

Here’s how it works (and why it matters way more than it sounds) ↓
Old voice search worked like this:

Speech → Text → Search.

If ASR misheard a single word, you got junk results.

Say “The Scream painting” → ASR hears “screen painting” → you get art tutorials instead of Munch.

S2R deletes that middle step completely.
S2R asks a different question.

Not “What did you say?”
But “What are you looking for?”

That’s a philosophical shift from transcription to understanding.
Read 9 tweets
Oct 6
I analyzed every single prompt in Anthropic's official library.

What I found will make you delete every "prompt engineering course" you bought.

Here's the framework they actually use:
First discovery: they're obsessed with XML tags.

Not markdown. Not JSON formatting. XML.

Why? Because Claude was trained to recognize structure through tags, not just content.

Look at how Anthropic writes prompts vs how everyone else does it:

Everyone else:

You are a legal analyst. Analyze this contract and identify risks.

Anthropic's way:

Legal analyst with 15 years of M&A experience


Analyze the following contract for potential legal risks



- Focus on liability clauses
- Flag ambiguous termination language
- Note jurisdiction conflicts


The difference? Claude can parse the structure before processing content. It knows exactly what each piece of information represents.Image
Second pattern: they separate thinking from output.

Most prompts mix everything together. Anthropic isolates the reasoning process.

Standard prompt:

Analyze this data and create a report.

Anthropic's structure:


First, analyze the data following these steps:
1. Identify trends
2. Note anomalies
3. Calculate key metrics



Then create a report with:
- Executive summary (3 sentences)
- Key findings (bullet points)
- Recommendations (numbered list)


This forces Claude to think before writing. The outputs are dramatically more structured and accurate.

I tested this on 50 prompts. Accuracy jumped from 73% to 91%.Image
Read 14 tweets
Sep 27
Fuck it.

I'm going to share the n8n workflow that turned my WhatsApp into Jarvis.

Send it any website link and it learns forever.

Here's how to build it (step by step guide 👇) Image
The workflow is brilliant. It starts with a WhatsApp trigger that catches both voice and text messages.

Voice notes get transcribed using OpenAI Whisper. Text goes straight through.

But here's the genius part - it uses a Switch node to route messages differently based on whether you're chatting or training it.
The "training mode" is what makes this feel like magic.

Send "Train: [website URL]" and watch it:

- Scrape the entire webpage
- Extract product name, price, description
- Store everything in a Google Sheet automatically
- Remember it forever

Your AI just learned something new in 3 seconds.
Read 9 tweets
Sep 24
This Stanford paper just proved that 90% of prompt engineering advice is wrong.

I spent 6 months testing every "expert" technique. Most of it is folklore.

Here's what actually works (backed by real research):
The biggest lie: "Be specific and detailed"

Stanford researchers tested 100,000 prompts across 12 different tasks.

Longer prompts performed WORSE 73% of the time.

The sweet spot? 15-25 tokens for simple tasks, 40-60 for complex reasoning. Image
"Few-shot examples always help" - Total bullshit.

MIT's recent study shows few-shot examples hurt performance on 60% of tasks.

Why? Models get anchored to your examples and miss edge cases.

Zero-shot with good instructions beats few-shot 8 times out of 10. Image
Read 16 tweets
Sep 23
Everyone says "be authentic" on LinkedIn.

Then they post the same recycled motivational garbage.

I've been using AI to write posts that sound more human than most humans.

10 prompts I use in Claude that got me 50K followers in 6 months:
1. Create a high-performing LinkedIn post

“You are a top-performing LinkedIn ghostwriter.
Write a single post (max 300 words) on [topic] that provides insight, tells a short story, and ends with a strong takeaway or CTA.”
2. Turn tweets into full LinkedIn posts

“Expand this tweet into a high-performing LinkedIn post.
Keep the tone professional but conversational. Add more depth, examples, and a clear lesson.”
→ [Paste tweet]
Read 13 tweets
Sep 22
Claude > ChatGPT
Claude > Grok
Claude > Gemini

But 99.9% of the users don't know how to get 100% accurate results from Claude.

To fix this you need to learn how to write prompts Claude.

Here's a complete guide on how to prompts for Claude using XML tags to get best results: Image
XML tags work because Claude was trained on tons of structured data.

When you wrap instructions in <tags>, Claude treats them as separate, weighted components instead of one messy blob.

Think of it like giving Claude a filing system for your request.
Basic structure that changes everything:

XML:

You are an expert data analyst


Analyze this dataset and find the top 3 insights



This is quarterly sales data from a SaaS company



- Insight 1: [finding]
- Insight 2: [finding]
- Insight 3: [finding]


vs

General prompt:

"Analyze this data and give me insights"
Read 12 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(