---
title: "Rabbits on Firetrucks Redux"
description: "Explore how AI image generation has evolved from unusable to professional-grade in 2 years, with key insights for startups using generative AI in business contexts."
categories: ["fundraising"]
keywords: ["generative AI","AI image generation","Coupa","Thoma Bravo","venture capital","SaaS","technology trends","image quality improvement"]
ai_summary: "Explore the evolution of AI image generation from unusable to professional-grade in just two years, with insights for startups."
date: 2025-04-21
lastmod: 2026-07-31
canonical_url: https://www.tomtunguz.com/rabbits-on-firetrucks-redux/
author: "Tomasz Tunguz"
---

Two years ago I published an [image of a rabbit on a fire truck](https://tomtunguz.com/gpt-stock-images/) to demonstrate how effective image generation was for blog posts. 

The conclusion at the time : the images were unusable. 

In 2025, the opposite is true - with one caveat.

**Prompt: Coupa & Thoma Bravo logos each in front of a cloud with a plus sign between them.**
 

<table style="width: 100%;">
  <thead>
    <tr>
      <th style="width: 50%; text-align: center;">2023</th>
      <th style="width: 50%; text-align: center;">2025</th>
    </tr>
  </thead>
  <tbody>
  <tr>
    <td style="width: 50%; text-align: center;"><img src="https://res.cloudinary.com/dzawgnnlr/image/upload/omjienz3hwdpeij79cks.png" alt="Image 1" style="max-width: 100%; height: auto;"></td>
    <td style="width: 50%; text-align: center;"><img src="https://res.cloudinary.com/dzawgnnlr/image/upload/xeds5noghmgs0m89yuee.jpg" alt="Image 2" style="max-width: 100%; height: auto;"></td>
  </tr>
  </tbody>
</table>


Two years has brought tremendous professionalism to new image. Corporate street art is now pixel-perfect professional. The clouds and the + sign are perfect, and the logos look the part. 

But! They are not the logos of either Coupa or Thoma Bravo. Perhaps sufficiently high-quality to be published under a less discerning editor or good enough to crop out the logos and replace the real ones by hand. 

**Prompt: a collection of crypto tokens on a cosmic background.**


<table style="width: 100%;">
  <thead>
    <tr>
      <th style="width: 50%; text-align: center;">2023</th>
      <th style="width: 50%; text-align: center;">2025</th>
    </tr>
  </thead>
  <tbody>
  <tr>
    <td style="width: 50%; text-align: center;"><img src="https://res.cloudinary.com/dzawgnnlr/image/upload/w33chrnvhgnpiee0gtxh.png" alt="Image 3" style="max-width: 100%; height: auto;"></td>
    <td style="width: 50%; text-align: center;"><img src="https://res.cloudinary.com/dzawgnnlr/image/upload/dntojsnip0tpxi2qqggc.jpg" alt="Image 4" style="max-width: 100%; height: auto;"></td>
  </tr>
  </tbody>
</table>


The crypto tokens are far more polished. They are different colors, consistent shapes. The Ethereum logo is repeated often. And the other logos are not recognizable tokens, but overall, again, a publishable image with enough professionalism & pizazz to grab readers' attention.

**Prompt: a Magic8 Ball with the year 2023 on the internal pyramid.**

<table style="width: 100%;">
  <thead>
    <tr>
      <th style="width: 50%; text-align: center;">2023</th>
      <th style="width: 50%; text-align: center;">2025</th>
    </tr>
  </thead>
  <tbody>
  <tr>
    <td style="width: 50%; text-align: center;"><img src="https://res.cloudinary.com/dzawgnnlr/image/upload/lzeaegozdslswj0sncxc.png" alt="Image 5" style="max-width: 100%; height: auto;"></td>
    <td style="width: 50%; text-align: center;"><img src="https://res.cloudinary.com/dzawgnnlr/image/upload/icde5ta4acdgjp4wgve5.jpg" alt="Image 6" style="max-width: 100%; height: auto;"></td>
  </tr>
  </tbody>
</table>

The Magic 8-Ball has made huge strides from the nonsensical 2023 iteration that struggled to generate coherent numbers  to a glass ball on a desk with a pyramid inside and the number of the year. 

It isn't quite the rendition I had imagined : peering through viewport at the bottom of a Magic 8-Ball with the year on the internal pyramid floating in the murky blue water, but again, reasonable enough to convey the idea.


Prompt: a rabbit on an Ikea fire truck.


<table style="width: 100%;">
  <thead>
    <tr>
      <th style="width: 50%; text-align: center;">2023</th>
      <th style="width: 50%; text-align: center;">2025</th>
    </tr>
  </thead>
  <tbody>
  <tr>
    <td style="width: 50%; text-align: center;"><img src="https://res.cloudinary.com/dzawgnnlr/image/upload/cmxblmdisq4zdv8tmkkk.png" alt="Image 7" style="max-width: 100%; height: auto;"></td>
    <td style="width: 50%; text-align: center;"><img src="https://res.cloudinary.com/dzawgnnlr/image/upload/dkaskfulqtad2ek7dzkt.jpg" alt="Image 8" style="max-width: 100%; height: auto;"></td>
  </tr>
  </tbody>
</table>

The only consumer image of the initial test, this image proved that consumer requests of things like a rabbit on a fire truck had a wide range of acceptable outcomes. 

That's clearly true ; however, the 2025 version is superior in almost every way. It's far more photo-realistic. The IKEA logo is emblazoned on the truck. The toy is properly sized. 

Within 24 months, image rendering technology has advanced to very nearly production quality with a single prompt.