Branding + AI strategies for first-generation founders.
Access AI Inner Circle →
Joined September 2017
- Tweets1.5K
- Following15
- Followers93
- Likes1K
Pinned Tweet
I combed through 10 years' worth of bookmarked websites and software to find the best tools for first-gen founders to grow their businesses.
🌪️ Sales Funnels
💸 Grants & Funding
🎁 Product Templates
📰 Copywriting Strategies
and more!
Easy to access.
divads.gumroad.com/l/khbux
3 Claude prompts that will 10x your AI output overnight:
1. "Ask me 3 questions before answering."
2. "Research online, then fact-check your findings."
3. "Grade your answer. Explain why it isn't 10/10. Rewrite until it is."
Save this.
If you're getting weak answers from ChatGPT or Claude, the problem is often your prompt.
Add these 3 instructions to every prompt:
✅ Ask me 3 questions first.
✅ Research online, then fact-check.
✅ Grade the result and rewrite until every category scores 10/10.
SAVE THIS.
How to make Claude review itself:
Add this line at the end of every chat👇
"Grade your answer on the 5 criteria that matter. Explain every weak score. Rewrite until every score is 10/10."
You'll spend less time fixing weak answers.
Don't treat Claude like a cheap wh*re. Take your time with it.
Make it ask YOU questions.
Make it verify YOUR THOUGHTS.
Make it improve itself.
3 prompt lines I use with Claude that 10x the output every time 👇
You'll never find the perfect prompt.
The bigger win is building a better process.
Good prompts help.
Good workflows compound.
I use this process inside my Conversion Copywriter.
It pushes Claude to think before writing, question weak copy, and improve until the message sells.
divads.gumroad.com/l/ydgjk
You can't expect AI to read your mind.
Start every prompt with:
"Before you answer, ask me 3 questions that would make your answer better."
Better context = better output.
In 2026, Anthropic documented a software pattern that fixed one of my biggest AI agent mistakes.
I front-loaded all my agents with a dozen tools and instructions from the start. Their decision-making and processing speeds dropped.
Here's what I changed👇
If you're building AI agents, apply this rule:
Small context.
Small toolset.
Load on demand.
Your agent should earn more complexity as the task unfolds, not carry all of it from the start.
Want to see this principle in action?
I used it to build my first AI employee.
Try it for yourself.
divads.gumroad.com/l/ydgjk