Why does Unity AI engine beat the competitors?
Last night I was explaining the current RickAndMortai AI TV engine to a mate.
We started talking about competitors.
I’m very convinced that Unity is the right tech stack to build an AI TV engine. TV in this sense doesn’t refer to just TV episodes - anything that resembles the idea of watching something incessantly like Reels, Youtube, Shorts, whatever.
What are the competitors building:
Python/Moviepy video generators for Instagram, Youtube, and Tiktoks.
Tiktok
Web video editors - Canva
Unity + a web app is fundamentally better. Here’s why:
Writing a script on the web and then generating an mp4 from it is easy. Building a performant video engine on the web is technically impossible at a certain level of difficulty - you can’t use the GPU, you probably build on OpenGL rather than the DOM thus reimplementing everything, Three.js and other 3D OpenGL engines are built for simple environments and lack the component GUI that allows you to compete better by generating new formats (ie. Joe Rogan interviews, Temple runner memes, etc.) at lower cost of development. Unity is performant and is easily deployable to the cloud, supports a host of formats for 2D, 3D, audio, animation, and more - and it has all the capabilities of a media environment you need - rendering text, animations, lighting, background images, and more.
Python/Moviepy cannot compete as a multimedia engine. It’s simply not the right problem-solution fit for generating things which are fun and easy. I’ve used it. It’s basically for subtitling and small video ideas and transitions.
Tiktok and other proprietary apps are probably the best competitor, mimetically. They may or may not adapt for the meme culture. I’ve written a LOT about what’s coming in the next 8yrs on this blog. The thing people forget is that Tiktok itself is a design/organisation - they have found PMF and are not going to change course very rapidly. The AI multimedia future I imagine involves AI agents on every platform, codeveloping lore with 4chan-like videoboards, pipelines of information and communities and objects, metaverse style. It sounds insane but it’s so different to Tiktok.
Canva has a video editor but it isn’t built for livestreaming or games. Livestreamed AI TV shows are definitely something that’s coming in one form or another.
What’s probably more of a risk/threat is the distribution platforms themselves. If you can generate engaging video content, people will copy the format easy. You can do that with some effort. What’s very defensible is the speed at which you can iterate on ideas until you achieve some sort of data defensibility - owning the assets and IP and the platform itself for building these things, ultimately even the distribution/hosting of the videos themselves, colocated with the interface. There’s a reason you tweet and read tweets in the same place.