Hello everyone, thanks for visiting my profile!

I make bots for people who enjoy romance and women-focused stories. Right now, I mostly create erotic content, but I’m working on building more engaging romance plots alongside it.

I also love worldbuilding, so most of my bots use a lot of tokens. Because of that, I recommend using DeepSeek or another proxy service for the best experience. A tutorial is linked at the bottom of my profile.

My current worlds are Azuria and Avalon.

Azuria (Modern Witches) - In a quiet American town, where the veil between the ordinary world and witch society runs dangerously thin. Beneath the halls of Azur High, rival covens hide in plain sight, weaving secrets, alliances, and vendettas while struggling to remain unseen.

Avalon (Dark Fantasy) - Avalon is a dark fantasy world where souls are currency, the Goddess demands constant sacrifice, and harems shape both love and power. Beautiful women, mythical creatures, and demi-humans fill its regions, torn between faith, war, and desire.

Completed (I can't add more tokens)

Current Plans

• Introduce faculty/mothers into the Azuria setting (Current Plans)

• Introduce a Sci-Fi setting (Long Term, TBD)

What I Could Produce

• Smut

• Vanilla

• Noncon

• Harem

• Demi Human

• NTR (Avoidable)

DeepSeek Tutorial

Just a quick link to other resources to help you use DeepSeek on Janitor AI. Also, because my bots work best with LLM models that can work with high token amounts

These three links were what I used to set mine up:
In-Depth Tutorial (On-Site) | Reddit Guide | YouTube Guide

Some things to note

When these tutorials mention linking Chutes or Taragon, consider not doing it. I just started using DeepSeek recently, but I've noticed a surge of past complaints linking directly to Chutes and Taragon. So for a more stable experience, simply use vanilla DeepSeek.

Personally, I use V3-0324, as it is a middle ground between R1 and V3. R1 is more chaotic, focusing on more negative aspects of characters (So, unless you want to REALLY freak out), and R3 is way more logical. V3-0324 is the best option IMO. I Also use the paid option, as I use up way more than 50 messages a day, and because I can afford a $10 top-up every now and then.

Lastly, when you have successfully linked the proxy, double-check your generation setting, and most importantly, refer back to the In-Depth, On-Site guide and set up your prompt. It is step 4. And also for your setting, as I said before, make sure your token count is 0, temp 0.8-9, and your context size is 32000.

It might take a while to set up, but it is way better than JLLM, and can totally upgrade your experience on this website. Of course, other options exist for proxies such as gemini and chatgpt, but I haven't tested those out.