The AI subscription fatigue is real. So I built a unified studio for Sora, Veo, and Flux.

Happy Holidays, everyone. 🎄

While the rest of the world is unboxing gifts, I’m doing something strictly forbidden on a holiday: looking at my SaaS burn rate.

It’s late 2025. The generative AI space has become paradoxically amazing and incredibly annoying.

Technically, we are living in a golden age. Sora 2 is mind-blowing. Veo 3 produces video quality we couldn't dream of a year ago. Flux has basically solved text-to-image.

But from a workflow perspective? It’s a mess.

The "Tab-Switching" Hell

Here is what my workflow looked like two months ago:

1. Generate a base image in Midjourney or Flux (Tab 1).
2. Download it, upscale it in another tool (Tab 2).
3. Upload it to Runway or Kling to animate it (Tab 3).
4. Realize the motion is weird, switch to Sora to retry (Tab 4).
5. Pay $30 here, $60 there, $20 there.

I realized I wasn't paying for "creativity"; I was paying for the privilege of fragmentation. I was burning $150/month just to have access to the models, even if I only used them sporadically.

Why I built MakeShot (The Logic, not the Pitch)

I built MakeShot.ai not because the world needs another "AI Wrapper"—I know, that word is a taboo here—but because we need a better Interface Layer.

If you are an indie hacker or a creator, you don't care which model does the job; you just want the shot.

Here is how I designed MakeShot to solve the fragmentation issue, and more importantly, the logic behind the features:

1. The "A/B Testing" of Creativity
We integrated the heavy hitters: Sora 2, Veo 3, Nano Banana, and Flux.
The real value here isn't just access; it's comparison.
In my own testing, I found that while Sora handles physics better, Veo often nails the cinematic lighting. Being able to run the same prompt through multiple engines in one UI saves massive amounts of trial-and-error time.

2. Solving the "Consistency" Problem
If you’ve tried making a story-driven video with AI, you know the pain: your main character looks like a different person in every shot.
I focused heavily on the Image-to-Video pipeline. You can upload reference images (up to 4) to "anchor" the generation.

• Dev note: This is crucial for anyone trying to build faceless YouTube channels or marketing assets. Randomness is fun for art, but terrible for business.

3. The Economics of Aggregation
Instead of five monthly subscriptions, I moved to a credit-based system that covers all models.
This was a tricky decision for me as a founder (MRR vs. Usage-based), but for the user, it just makes sense. You shouldn't have to pay a full subscription to Veo if you only need it for 3 clips a month.

A Christmas "Stress Test"

To prove to myself that this workflow is actually viable, I spent this morning generating assets for a side project using only MakeShot.

I generated a consistent cyberpunk Santa character using Flux, then ran it through Veo 3 for motion. The friction of "download-upload-login" was gone. It felt less like using disparate tools and more like working in a cohesive studio.

Feedback Request

I’m putting this out here because the Indie Hackers community sees through marketing fluff better than anyone.

I’m curious about your take on the "All-in-One" vs. "Best-of-Breed" debate in AI.

• Do you prefer managing separate subscriptions to get the native features of each platform?
• Or are you ready for an aggregated layer that simplifies the mess?

I’ll be hanging around the comments section between Christmas meals. If you have any technical questions on how we handle the model integrations or the prompt engineering backend, ask away.

Check it out here: MakeShot.ai

Merry Christmas and happy building! 🚀