Hope you enjoy this video! Here's an idea to take this workflow to the next level! You can add a RAG agent using a custom knowledge base in this flow. Check out my RAG Agent tutorial to learn how. th-cam.com/video/UeFi5oV9UpY/w-d-xo.html
Great video! I learned more about n8n in the first 10 minutes of this video than I did watching a bunch of other tutorial videos on youtube. Now I'm going to binge-watch your other videos 😊
nice guide. Though I think it would have been worth mentioning in the video that this would forward all incoming emails to OpenAI and would also upload them to the n8n server, because you know a bunch of people will do this without realizing that.
Solution: self-host n8n and open-source models instead. Most if not all TH-camrs are "promoting" OpenAI. Many of them have good knowledge/content and we can learn from them, sad thing is that they are so stuck on ChatGPT.
Good tutorial for sure. On a note to related to the subject: I would use the email header to query a security service API of choice to validate the email's origin etc., even though Google does that. You can also do that with any links in the email body. Hence, responding to spam email is a bad idea.
Fantabulous as always. We know this workflow exists but when you do a tutorial, its a joy to watch.. there is always some bits to learn. Keep it coming !! One Question. When you scrub thru all the emails, will it not make all the email as read? on the false node, can we not change it back to unread?
Hello Leon thank you for your help. One big question because you have worked with the different tools. Which one do you prefer Flowise, n8n... ? thnak you a lot
I watch all your videos Leon and love them all! But it is confusing to work out the ins and outs of each solution : Flowise, VectorShift, n8n, Chatling.. It might interesting doing 1. comparison videos between these products to understand which one is best for specific requirements. And 2. how can these tools potentially integrate with each other. Just a suggestion from a fan of yours ! ;-)
@@leonvanzyl Yes please! And there are interesting videos on integrations between Flowise and other tools, or VectorShift and VAPI. I am unclear and confused if it is possible to integrate these tools you present or if they're just competitors. VS to VAPI example: th-cam.com/video/6AABOYkt_Wc/w-d-xo.htmlsi=56TestixtAHHGe7Q
Greetings Leo, I would like to know if you could record a video showing us how we can create a solution where multiple agents can perform a specific task simultaneously and in sync. Also, I want to know if it's possible to use the new Swarm framework from OpenAI with n8n. I would really appreciate your response. I trust your professional opinion a lot. A big hug, you’re a legend!
I'll look into this for you and create a video if it's possible. What would be the use case? I don't think n8n supports the swarm framework. I believe it uses Langchain under the hood. I'll look into it
@@leonvanzyl Hi Leo, thank you so much for your response and willingness to help! I’ve been thinking about an interesting use case that would likely catch the attention of the digital marketing community: You could showcase how to implement a solution where multiple agents work together to analyze a Google Ads campaign. The idea is that each agent handles a specific task and passes the processed data to the next agent to complete a thorough analysis. For example: 1. One agent extracts performance data from Google Ads (clicks, impressions, CPC, etc.). 2. Another agent analyzes this data to identify low-performing keywords or audiences. 3. A third agent evaluates the ad creatives (images and text) to detect successful patterns. 4. Another agent suggests optimizations such as budget adjustments or audience segmentation. 5. Finally, an agent generates a visual report that can be automatically shared with the client or team. This would be incredibly useful for those of us managing ad campaigns, as it automates much of the analytical process.
Love your video and this workflow! I set it up to test some scenarios with my personal email. Funnily enough I received draft responses to my draft responses. Any suggestions on ensuring it only makes a single draft response? Each log said succeeded but I don't know why it ran so many times.
Are you running it manually? The workflow will always process the last email when running manually. When you activate the workflow it should only process the email once..
@@leonvanzyl Hi Leon, I had it active (not running manually). It did this on a few of my emails. I may have missed the step but chatgpt'd that I need to set a Drafts filter on Label ID's. :)
Leon, I do follow you, but your not really advancing very much in last few months, I look to see what you video is about but I think a lot of your followers should be well ahead of most of what you show. Might be worth trying something different, more intricate, advanced if you are to keep your following. Having said that, if your videos are aimed at the beginner, change nothing, because they are great for them. Just my thoughts. 🙂
Hey there, thank you for the constructive advice. The n8n series is still building up in complexity, but I do have more advanced videos on things like agentflows in Flowise. If you're into coding then I think my upcoming LangGraph series might interest you. It's always a difficult balance to try and keep people of all levels engaged 😊. Is there anything you would like me to cover at your level?
@@leonvanzyl I love langgraph I am patiently waiting on them releasing the linux version of langgraph-studio. Thanks for the reply. I wasn't being rude, I think your channel is excellent, but to keep the following you have worked hard for, thats what my suggestion was about. Thanks
Hope you enjoy this video!
Here's an idea to take this workflow to the next level! You can add a RAG agent using a custom knowledge base in this flow.
Check out my RAG Agent tutorial to learn how.
th-cam.com/video/UeFi5oV9UpY/w-d-xo.html
Great video! I learned more about n8n in the first 10 minutes of this video than I did watching a bunch of other tutorial videos on youtube. Now I'm going to binge-watch your other videos 😊
Welcome to my channel 😁
nice guide. Though I think it would have been worth mentioning in the video that this would forward all incoming emails to OpenAI and would also upload them to the n8n server, because you know a bunch of people will do this without realizing that.
Well, you could just run this locally (or self host it) and use Ollama as the chat model 😁.
I did this in a previous video.
Solution: self-host n8n and open-source models instead. Most if not all TH-camrs are "promoting" OpenAI.
Many of them have good knowledge/content and we can learn from them, sad thing is that they are so stuck on ChatGPT.
Good tutorial for sure.
On a note to related to the subject:
I would use the email header to query a security service API of choice to validate the email's origin etc., even though Google does that. You can also do that with any links in the email body.
Hence, responding to spam email is a bad idea.
Excellent idea 💡
Matteo is a community treasure.
Fantabulous as always. We know this workflow exists but when you do a tutorial, its a joy to watch.. there is always some bits to learn. Keep it coming !! One Question. When you scrub thru all the emails, will it not make all the email as read? on the false node, can we not change it back to unread?
You're welcome 🤗.
The workflow actually didn't change the emails to read for me.
Hello Leon thank you for your help. One big question because you have worked with the different tools. Which one do you prefer Flowise, n8n... ? thnak you a lot
Both.
I would recommend learning Flowise along with an automation tool like n8n or Make.
Awesome Video Leon
Thank you
i want somethign locally ran and or hosted. look forward to that.
I watch all your videos Leon and love them all! But it is confusing to work out the ins and outs of each solution : Flowise, VectorShift, n8n, Chatling.. It might interesting doing 1. comparison videos between these products to understand which one is best for specific requirements. And 2. how can these tools potentially integrate with each other. Just a suggestion from a fan of yours ! ;-)
Great suggestion.
I really do need to make a comparison video.
@@leonvanzyl Yes please! And there are interesting videos on integrations between Flowise and other tools, or VectorShift and VAPI. I am unclear and confused if it is possible to integrate these tools you present or if they're just competitors. VS to VAPI example: th-cam.com/video/6AABOYkt_Wc/w-d-xo.htmlsi=56TestixtAHHGe7Q
Amazing! Thanks -
You're welcome 😁
Greetings Leo, I would like to know if you could record a video showing us how we can create a solution where multiple agents can perform a specific task simultaneously and in sync. Also, I want to know if it's possible to use the new Swarm framework from OpenAI with n8n. I would really appreciate your response. I trust your professional opinion a lot. A big hug, you’re a legend!
I'll look into this for you and create a video if it's possible. What would be the use case?
I don't think n8n supports the swarm framework. I believe it uses Langchain under the hood. I'll look into it
@@leonvanzyl Hi Leo, thank you so much for your response and willingness to help! I’ve been thinking about an interesting use case that would likely catch the attention of the digital marketing community:
You could showcase how to implement a solution where multiple agents work together to analyze a Google Ads campaign. The idea is that each agent handles a specific task and passes the processed data to the next agent to complete a thorough analysis.
For example:
1. One agent extracts performance data from Google Ads (clicks, impressions, CPC, etc.).
2. Another agent analyzes this data to identify low-performing keywords or audiences.
3. A third agent evaluates the ad creatives (images and text) to detect successful patterns.
4. Another agent suggests optimizations such as budget adjustments or audience segmentation.
5. Finally, an agent generates a visual report that can be automatically shared with the client or team.
This would be incredibly useful for those of us managing ad campaigns, as it automates much of the analytical process.
Love your video and this workflow! I set it up to test some scenarios with my personal email. Funnily enough I received draft responses to my draft responses. Any suggestions on ensuring it only makes a single draft response? Each log said succeeded but I don't know why it ran so many times.
Are you running it manually? The workflow will always process the last email when running manually.
When you activate the workflow it should only process the email once..
@@leonvanzyl Hi Leon, I had it active (not running manually). It did this on a few of my emails. I may have missed the step but chatgpt'd that I need to set a Drafts filter on Label ID's. :)
Can you do this with the self hosted version? the callback URI will never be acessible, right?
This works with local self hosting too 👍
@@leonvanzyl how do you deal with the AI callback?
Please çontinúe with more examples, very exciting!
Will do!
Any ideas?
Insane
Leon, I do follow you, but your not really advancing very much in last few months, I look to see what you video is about but I think a lot of your followers should be well ahead of most of what you show. Might be worth trying something different, more intricate, advanced if you are to keep your following. Having said that, if your videos are aimed at the beginner, change nothing, because they are great for them.
Just my thoughts. 🙂
Hey there, thank you for the constructive advice.
The n8n series is still building up in complexity, but I do have more advanced videos on things like agentflows in Flowise.
If you're into coding then I think my upcoming LangGraph series might interest you.
It's always a difficult balance to try and keep people of all levels engaged 😊.
Is there anything you would like me to cover at your level?
@@leonvanzyl I love langgraph I am patiently waiting on them releasing the linux version of langgraph-studio.
Thanks for the reply. I wasn't being rude, I think your channel is excellent, but to keep the following you have worked hard for, thats what my suggestion was about.
Thanks