Best Self-Hosting Options for Scaling an Agentic RAG SaaS?

No Code AI
1

First post here! I’ve been working on a SaaS solution based on techniques I personally used to navigate a real-world issue. So far, I’ve built a significant portion of it using Lovable.dev and have integrated it with:

  • GitHub
  • Supabase
  • Netlify

Now, I want to connect n8n to implement an Agentic RAG system. My concern is that once it’s fully operational, the costs of n8n, AI interactions, and document processing—along with the AI chat feature—could become expensive as user adoption scales.

I’ve considered self-hosting n8n and the LLM on a cloud service like DigitalOcean, but setting all of that up is beyond my current capabilities. Based on my calculations, even the lowest paid plan on n8n would likely only support around 8 users, which isn’t sustainable as I scale.

Looking for Advice on Scaling

  1. What are the best self-hosting options for reducing long-term costs?
  2. Are there more efficient architectures I should consider?
  3. If anyone has experience working on a project like this, I’d love to connect with an experienced partner who enjoys this type of work.

Would appreciate any insights from those who’ve tackled similar challenges!

1 Like
2

Love what you’re doing Rob!

For really cheap and powerful cloud hosting for n8n I’d take a look at Hostingr VPS! And in general, even the cheaper options for DigitalOcean and other platforms will scale pretty well with n8n. I’m curious what research you found to make you think it would only scale to 8 users?

3

Thanks for the feedback, Cole!

For the 8 person limit statement, I was going by the capabilities of the Starter package for n8n and it says it only handles 2500 executions a month. I figured if I cap each user at 10 executions a day, for 30 days a month (on average), then that’s 2400 so I can only manage about 8 users per month.

This was my thought process before seeing your most recent video about self hosting the entire stack through a cloud provider instead of using the paid version. Both sites looked easy to work with for cloud hosting, but I am leaning towards DigitalOcean for scalability. I’ll start with a cheaper plan until I get a higher number of users.

My next question is, even though I am using the same agentic RAG workflow for different users, it should double check the id assigned for every user and only store their conversation into a separate memory, correct?