Last Updated: September 2026
AI Declaration
This is a comprehensive and truthful declaration of what AI products (generative or otherwise) I have used in the past and currently still use. This page will be updated if anything changes.
Commercial Writing
I do not use generative AI in my writing process. In all commercial writing, I write and edit all content myself.
My first draft is normally completed by "writing with my mouth" using the Letterly app. I find it easier to dictate and think than to think and write at the same time.
The Letterly app is useful because it is better at picking up spelling than most dictation apps, and can read fantasy words from a custom dictionary that I provide.
The notes from that first draft go into LibreOffice, which is my current preferred word processor, and I use LanguageTool for a spell checker and grammar editor.
This applies to both non-fiction writing, like "World Building Blueprints," as well as fiction writing, like "Sangwheel Chronicles."
I do not use generative AI images in any commercial content. All images (interior art or cover art) are either created by human artists who, to the best of my knowledge, do not use generative AI, or is purchased from Shutterstock from artists who, to the best of my knowledge and per their declaration on the images, have not used generative AI to create those images.
I do not use generative AI in the creation of audiobooks. I narrate my audio books myself and edit the sound myself. I do use the CleanVoice app to reduce my breaths and remove mouth sounds, as this cuts down on a lot of manual work I used to do in the sound editing process.
I did create two AI narrated audio books, the two erotica novellas written under the name Talia Tethers. This was done as an experiment to see how it worked and I couldn't test without allowing Google Play to distribute the audio books. They are clearly marked as AI and having done the experiment with these books, I'm happy to say I don't recommend doing this, AI narration is crap and the process of setting the books up not much easier than just reading them into a microphone.
YouTube Videos
I do not use generative AI in the creation of my YouTube Videos for either voice, b-roll, main video, or images. I do not use generative AI to create images for thumbnails.
I did go through a phase of using generative AI images for my thumbnails on YouTube and as some of the B-roll images in some of the older videos. I am in the process of replacing these thumbnails (the videos can't be replaced), but it is slow going as it is rather a lot of videos.
I used to take the stance that using generative AI on free content was reasonable (as in, using it for thumbnails), but I've changed my mind. I just don't want to see generative AI slop anymore.
I will sometimes use tools that allow me to blend images together rather than blending them by hand (images I have copyright to). For example: I might transplant my face on top of a elf's body or something like that. This, however, is extremely uncommon and requires a very specific idea for a thumbnail.
I do not create the video scripts using generative AI. I use LLMs as search engines to find web based content on the topic that I am researching. I read the sources found from the web, open access papers, and books with my own eyes. I listen to podcasts / watch videos with my own ears.
Once I have a script outline, I use Letterly to dictate the first version of the script. I then edit the script and uploaded it in DOCX format to my teleprompter.
I use an app called CleanVoice to reduce the volume of my breaths on the sound track. I used to do that by hand and, as anyone whose done sound editing will tell you, breath reduction is a very painful and exacting task. CleanVoice reduces the time that I spend on video editing by a great deal as it removes the grunt work of finding each breath and reducing its volume.
After the video is uploaded, I do take the automatically generated script of the video, and use an LLM to clean up the grammar and spelling. This also is a task I used to do manually, and it can be very slow because of the fantasy words and names. Using an LLM for this base cleanup frees up roughly an hour of my time per video.
And that's it. That's my workflow :)
