Prompt engineering is not always about writing better inputs.
You can make the AI fix its own answers. It’s called Self-Refine.
Here’s how this technique improves your output quality by 20% 🧵:
1/ Self-Refine is a prompting technique where a single LLM plays 3 roles:
1️⃣ Generates a response
2️⃣ Critiques its own response
3️⃣ Improves it based on that feedback
This repeats until the model decides it’s “good enough.”
Let’s break it down with real-world examples 👇
2/ 🎾 Example 1: User Inquiry (Table Tennis)
🟢 Prompt:
I need you to respond to this user message in a helpful way:
User: "I am interested in playing table tennis."
Response:
"I'm sure it's a great way to socialize, stay active and have fun. If you're looking to get started, you might want to find a local club or community center that offers tables and perhaps lessons for beginners."
I turned ChatGPT o3 into my stock trading assistant.
It's an edge 99% of traders still don’t know exists.
Here are 10 INSANELY powerful prompts I use 🧵:
1/ Advanced Technical Analysis
Prompt: Technical Pattern Recognition Specialist Analyze the stock chart for [TICKER] over the past [TIMEFRAME] using advanced pattern recognition. Required: Stock symbol Required: Duration for analysis Optional: Candlestick/Line/OHLC (default: Candlestick)
1. Identify all significant chart patterns (Head & Shoulders, Cup & Handle, etc.)
2. Calculate precise support/resistance levels including hidden levels
3. Plot trendlines with mathematical precision
4. Analyze volume patterns and what they indicate about smart money movement
5. Apply Fibonacci retracement/extension to identify key price targets
6. Evaluate momentum with multiple timeframe RSI, MACD, and Stochastic analysis
7. Calculate volatility metrics and expected price ranges
2/ Market Catalyst Impact Analyzer
Prompt: Event-Driven Trading Strategist Analyze how [UPCOMING EVENT] will impact [TICKER/SECTOR] across different timeframes. Required: Specific market catalyst (Fed meeting, product launch, etc.) Required: Stock symbol or market sector Optional: Currently long/short/none (default: none)
1. Research historical stock/sector reactions to similar events (exact % moves)
2. Determine statistical significance of these reactions
3. Analyze current market positioning data (COT reports, fund flows, etc.)
4. Calculate potential impact magnitudes (best/base/worst case scenarios)
5. Design optimal pre-event positioning strategy
6. Create post-event reaction playbook for each scenario
7. Recommend specific options strategies to capitalize on volatility changes
You can turn any logo into stunning 3D art — no design skills needed.
No Photoshop. No Midjourney. Just ChatGPT.
Upload your logo, describe your vision… and it generates the image for you.
Here are 13 mind-blowing prompts 🧵👇:
1/ ChatGPT can now generate 3D logo renders with textures, lighting, shadows — all from a simple prompt.
You just:
→ Upload your logo
→ Tell it what you want to turn it into
→ Get a fully rendered image
Here are the prompts👇:
2/ Fluffy 3D Logo
Prompt I used:
"Transform [BRAND] logo into 3D fluffy object with soft fur texture. Maintain exact brand colors throughout the plush material. Include realistic hair fibers with subtle movement, variable fur length, and gentle shadowing between strands. Render on transparent background. Studio lighting creating soft highlights on fur tips. Photorealistic rendering with fine detail in fiber structure."
But for tough business decisions, one answer isn’t enough.
Tree of Thoughts makes AI think in branches, debate outcomes, and explore every path.
Here’s why it changes everything 🧵:
1/ What is Tree of Thoughts (ToT)?
It's a prompting technique that helps AI explore multiple solution paths simultaneously, evaluating each step before proceeding.
Think of it as letting AI "think ahead" like a chess player instead of giving you the first answer that comes to mind.
2/ Why entrepreneurs NEED this:
- Complex decisions with many variables
- Strategic planning requiring foresight
- Problem-solving when straightforward approaches fail
- Decision making with multiple possible outcomes