Look forward to this roll out @jan - Will be interesting know more about the prompting methods required to get the best output. What would your recommended approach be regarding prompt method?
1 Like
Great news! Thxs…
1 Like
Thanks for sharing the information, I would like to sign up for the beta version.
1 Like
Hi @jan - it is really a great AI Agent tool, and i am really exicted to have this prompt features in the n8n so that people will be using it more and it will making there work easier and effective.
@jan - try to make this pricing like on demand type for example AWS EC2 so how much Ai prompt will be used it will chagered after some credits.
2 Likes
Sounds amazing, I really hope the self-hosted version will have the option to just use a API key for that.
2 Likes
Amazing 
1 Like
great news that we can use AI to help with building flows. But please also support running local models
1 Like
To To good news - from email in the morning - this is huge!
Thank you @jan for this great info and because it will be in free version it opens up opportunities for rapid prototyping and huge feedback of a tool! I can’t wait to test it when the beta update arrives.
Like @solomon I’m also positively surprised that you didn’t hide IP and open it to for free - n8n making standards and ripples hahaha 
1 Like
that is amazing! I can’t wait to try it already. As some other people already mentioned the way to go would be to put your own LLMs credentials as for agents unless it is a fine tuned model you made, in which case you could also publish and bill it through openrouter and similar I think
2 Likes
Currently, my best experience using AI to build n8n workflows is with Comet (Perplexity), as it understands the context of n8n flows, expressions, etc., very well 
(not a sponsored comment 
).
4 Likes
Awesome feature, can’t wait.
1 Like
Thanks, I will also check this out
2 Likes
That sounds amazing 
1 Like
This is an exciting feature; for the self-hosted, it would be amazing if BYOAI were an option to power this feature. This could alleviate the cost concern from n8n’s perspective.
1 Like
Thanks for this work, and thanks for considering the self-hosted community! It really warms my heart 
Don’t overextend yourselves either! In any case, I can’t wait to try it out!
2 Likes
We really should be able to use our own API credentials and/or LLM if ran locally. Does not make sense for n8n to take on the cost and pass it down to the users who are likely already paying for a LLM.
If n8n proceeds with embedding a model they pay for, in turn user pays for, then it should be a “add-on” so we are not forced to pay more for a feature we don’t really need/intend to use.
3 Likes
Thanks for including us self/locally hosted folks!
1 Like
It’s great @jan and the team, can be a real game changer 
180k here we are!
1 Like
Hi @jan
First of all, thank you to you and the whole n8n team for building such a fantastic platform. I’ve been using n8n for a while now, and I truly appreciate the power and flexibility it gives me. It’s clear how much passion and effort goes into the product, and it has become an important tool in my daily work.
I saw your announcement about the upcoming AI Workflow Builder. I understand why this feature is exciting and how it can help many new users get started more easily. At the same time, I wanted to share some feedback from the perspective of someone who enjoys building workflows step by step. For me, the joy of n8n lies exactly in designing these flows myself — that’s what makes the platform so unique compared to others.
Where I (and many other power users) currently struggle most is with the AI Agent Node and GPT-5 integration. At the moment it’s still difficult to use GPT-5 in a stable way, and this limits what we can build. A few examples:
- Reliable support for the new Responses API is missing.
- Parameters sometimes don’t behave as expected, which causes errors.
- Structured JSON outputs are not always easy to enforce.
These issues make it hard to use the AI Agent Node in production, even though it could be one of the most powerful features of n8n.
My personal wish would be to see the Agent Node + GPT-5 integration prioritized and improved before too much focus goes to the Workflow Builder. Strong Agent capabilities would really empower those of us who are building advanced automations, and I believe it would strengthen n8n’s core identity as a flexible automation platform.
Thanks again for everything you and your team are doing — I share this only because I care about the direction of the product and want to see it succeed for both new and advanced users.
Cheers,
Fabian
2 Likes
That’s pretty damn cool and the natural evolution
1 Like