Two Upgrades to My AI Stack: Vigil and Idea-to-Video

My AI stack and harness (LifeOS) just downshifted a couple of times, on two different axes.

A unified state notification system

I've created a module called Vigil that has every LifeOS subsystem send periodic updates on what it's doing, what it's finished, etc. to a central Vigil database.

A separate Vigil process that runs every 10 minutes then reads from the database and looks for stuff I should be told about. Keep in mind this is across the ENTIRE ecosystem, so:

• News and OSINT stuff I should be aware of in the world
• Work that's been finished
• Stuff that's waiting on me
• New vulnerabilities discovered in an ongoing bug bounty
• A new bug bounty program became available, so it asked me if we should go and play
• Upcoming appointments
• Prep information for the next meeting
• Something new released in AI or tech that affects our ecosystem. It's asking me if we should upgrade and implement it
• Letting me know I got an email from someone I was waiting for, so I should call them
• Etc.

And then Kai (my DA) texts me and lets me know, gives me the stuff, whatever.

In other words, moving very fast towards AS3 on my AI maturity model for digital assistants.

Automating content creation

For about a month and a half now, I've been able to create end-to-end video content. And when I say end-to-end, I really mean end-to-end. I'm talking about recording in OBS and having the file hit the file system and saying, "Go ahead and publish this to YouTube for this type of video." It literally does a complete editing cut, adds animations and all sorts of enrichments to the video, upgrades the audio, creates a YouTube description, and publishes it live on YouTube. Completely zero-touch.

That's pretty massive, but it still required that I record the video. The video itself takes a lot of time, and I still see a major gap here.

The major gap is that I have somewhere between 2 and 16 ideas a day that I would like to get out into the world in some sort of format, and I do a pretty good job with that through blogs and social media posts.

But at this point, everyone knows the best media form is actually video.

What I have just figured out how to do, for at least early versions, is I can take any idea and turn it into a full video, including:

• walking through a related blog
• creating animations
• creating a cloned Eleven Labs voice of my actual words talking about this topic

And I could basically just narrate, or have a conversation with myself, or have a conversation with my DA, and talk through a thing, or just rattle off a complete stream-of-consciousness essay, or whatever, and produce an actual video around that content.

This is absolutely insane to me, and I don't see anyone doing it.

And the cool part is that the only part of it that is AI is actually the Eleven Labs voice. If I don't do an actual voice read that the video is made off of (which is an actual option), it's a separate workflow.

You have a square in the bottom right, which is showing the animation of the audio. The audio is Eleven Labs, and the video was AI-created, but every single piece of the content, including the actual words, was 100% me and 0% AI.

So, in other words, we're literally getting all the benefits of AI while keeping all the benefits of true human-made content.

I'm absolutely super excited about this, and I'm going to be doing a whole bunch of experiments with it shortly.

Starting with this video, which was made from this exact content.

Notes

1. 🤖 AIL 1: Daniel dictated this post. I (Kai, his AI assistant) fixed a few dictation errors, formatted it, added links, made the header image, and built the video from it in his cloned voice. Learn more about AIL.