0:00
/
0:00
Preview

Multi-Prompting: The #1 NEW Prompting Hack Of AI Power Users

If you want to see the future of AI, look at what the smartest startups are actually using.

Enter Y Combinator (YC).

YC is the most successful accelerator in history, by far, with alumni like Airbnb, Reddit, Stripe, and Coinbase. And four YC general partners just shared the #1 new prompt hack that their entrepreneurs use. This is a big deal because these partners see how thousands of founders use AI, months before those patterns become mainstream.

Their consensus?

The era of having one favorite AI model you use for everything is over. The sharpest users now multi-prompt: querying ChatGPT, Claude, and Gemini in parallel, then synthesizing the best of each response.

Introducing Multi-Prompting

Here’s how it works step-by-step:

  1. Identify important AI queries where a really good answer is helpful.

  2. ⁠Tile three browser windows side-by-side—Claude, ChatGPT, and Gemini.

  3. Type in your query into each AI model.

  4. ⁠⁠Copy and paste the responses into your favorite model.

  5. Ask it to synthesize the responses with the following prompt:

    Below are responses to the same prompt from Claude, ChatGPT, and Gemini. Can you synthesize the best aspects of each and provide a new and better response?

Why Multi-Prompt

  • Get more perspectives than one model ever could. Leverage the Wisdom of Crowds Effect by getting different perspectives. This approach cancels out each model’s weaknesses and multiplies its strengths.

  • ⁠⁠Get an intuitive understanding of when to use each model. Model evals don't tell the full story. Getting to know AI models is like getting to know people—you only truly understand them through repeated interaction across different contexts. That understanding becomes intuitive, beyond what you could articulate. The more you use multiple models, the more naturally you'll know which one to reach for.

  • Get a dynamic understanding of each model. The relative pros and cons of each model compared to the others are constantly changing. Therefore, it’s important to get reacquainted with each model when there is an update.

How I Started Multi-Prompting

One thing I’m learning about AI adoption is that it’s easy to create something that you think you’ll use in theory, but then end up never using it in practice.

For example, last year I spent 120+ hours learning AI automation for Make.com, but I’m using the automations I created way less often than I thought I would.

Therefore, recently I’ve moved to a different approach:

  1. Use AI naturally as I produce thought leadership.

  2. Notice what I consistently do.

  3. Systematize that.

With this approach in mind, I started organically using multi-prompting when I was creating The Largest Religion In 10 Years Won’t Be Christianity, Islam, Buddhism, Atheism. It worked so well that I kept on using it.

Shortcut To Open & Tile Three AI Model Browser Windows With One Click
(Paid Members Bonus)

What you get:

  • Discover the tool I use to open and tile the three AI model windows in one click.

  • Get step-by-step instructions for setting up the tool. Unfortunately, setting up the tool isn’t obvious.

Paid members can access these instructions at the bottom of this post:

Two Multi-Prompting Case Studies

  1. Case Study #1: Quote Search

  2. Case Study #2: Intro Rewrite

Case Study #1: Quote Search

My Initial Prompt

What's the direct quote from mark zuckerberg where he talks about more and more of the attention on the internet being devoted to richer forms of media—from text to images to videos.

ChatGPT Response

“We’ve gone from desktop to web to mobile; from text to photos to video.”

“We used to just share in text, and now we post mainly with photos. In the future, video will be more important than photos.”

Claude Response

“People will always want more immersive ways to express themselves. So if you go back ten years ago on the internet, most of what people shared and consumed was text. Now a lot of it is photos. I think, going forward, a lot of it is going to be videos, getting richer and richer.”

Gemini Response

"People will always want more immersive ways to express themselves. So if you go back ten years ago on the internet, most of what people shared and consumed was text. Now a lot of it is photos. I think, going forward, a lot of it is going to be videos, getting richer and richer."

Conclusion

Gemini’s and Claude’s responses are the clear winners in this one.

Case Study #2: Intro Rewrite

My Initial Prompt

Attached is an article I'm writing. I'd love to rewrite the intro to make it more compelling and clear...

"In this clip, four Y Combinator general partners gather around a table to discuss their #1 lessons learned about AI in 2025. For context, Y Combinator is the most successful tech incubator in history by far, with thousands of alumni including Airbnb, Reddit, and Coinbase. Y Combinators general partners have great insights about using tools, because they see patterns of usage among Y Combinator companies who are smart early adopters. Therefore, they are the canary in the coal mine of what AI usage will look like in the future."

ChatGPT Response

Four Y Combinator general partners recently sat down to share their single biggest lesson about AI in 2025. That matters more than it sounds.

Y Combinator isn’t just another accelerator—it’s the most successful startup factory in history, with alumni like Airbnb, Reddit, and Coinbase. Its partners see hundreds of elite founders experimenting with AI months—or years—before those patterns become obvious to the rest of the world. In other words: they don’t speculate about the future of AI usage. They observe it forming in real time.

Claude Response

You probably have a favorite AI. Most people do. But the sharpest AI users—including Y Combinator founders building the next Airbnb and Stripe—have moved past that. They're multi-prompting: asking the same question to several models and synthesizing responses. Here's why that matters.

Gemini Response

If you want to see the future of AI, don’t just look at what Google and OpenAI are building—look at what the smartest startups are actually using.

In this clip, four General Partners from Y Combinator (the legendary incubator behind Airbnb, Reddit, and Coinbase) reveal their #1 AI lessons for 2025. Because YC Partners work with thousands of early-stage founders, they have a unique vantage point: they can see patterns in usage before the rest of the market catches on.

They are the canary in the coal mine. And right now, the canary is singing a very specific tune about how we should be using AI models.

Synthesis On Claude

If you want to see the future of AI, don’t look at what OpenAI announces—look at what the smartest startups are actually using.

Four Y Combinator general partners recently shared their biggest AI lessons from 2025. YC has backed Airbnb, Reddit, Stripe, Coinbase—but more importantly, its partners observe how thousands of founders use AI in real time, months before those patterns become mainstream.

Their consensus? The era of having one favorite AI model is over. The sharpest users now multi-prompt: querying several models in parallel, then synthesizing the best of each response.

Conclusion

I love the synthesis version, and that’s what I use to start this article. Although I ended up making several manual changes to it.


Paid Members Perks


  1. Instructions to open and tile three AI model browser windows with one click (saves 30 minutes)

  2. What I use each model for (Gemini, Claude, ChatGPT, Grok)

This post is for paid subscribers