IMPORTANT: Import the workflows by opening up a new workflow, click the three dots in the top right, and select "Import from file". You must make sure when you are setting up the workflows as a tool for the main agent that you are choosing the correct workflow within your n8n environment.
I really wonder how a system like this can be integrated inside a website, in the user's dashboard for example, I'd love to see something like this, or at least for you to outline the steps needed to do this. Thanks for the great content!
Great one I really like the error handling from the agent area. This really makes it even more of a challenge to not consider an end for a truly agentic framework. Btw you showed your Tavily token in the video.
Thanks! I love the way you can have them communicate, it really makes me think about the different ways we can make workflows even more autonomous. Don’t worry, I always delete any keys after I publish a video :)
Hi Nate, Thank you for your content-it’s very useful! I really appreciate the videos and the n8n flows you’ve shared. They’re awesome! It would be great to see more common scenarios from a daily office job, maybe something involving a CRM API agent. Keep up the great work!
lol. You completely caught me off guard with the Michael Scott @ Dundler Mifflin email. Keep up the awesome content! I'd like to see more custom nodes/js/python/hugging face models etc. I find all the system level stuff difficult to learn, n8n is helping bridge the gap for me.
Please make a video on how you go about solving errors with the ai. For example, i tried making a meeting scheduling agent to use google Calendar and have spent hours trying to get it to stop messing up the available times. It says that completely empty days are full and sometimes ignored if some things are already scheduled, and randomly says available times. For example, on a fully empty day, it might say: were available from 9-10:15 11-3 and 3-4:30.
With AI agents, troubleshooting really comes down to 2 things: 1) repetitions and 2) trial and error. Especially when you’re building something from scratch so many things can go wrong, but understanding why they went wrong is what helps you learn and prevent them in the future!
To me it sounds like you should check if it has the right source data, and then if the context is too long - as you get closer to the content length limit some of that context will be forgotten / hallucinated about. Format and iterating over slices of data, or utilizing RAG approaches become necessary
How can I make a local host n8n version of this? I’ve been wanting to give commands solely to my Laptop and every once a while connect to the internet but still connected to my local server ? Is that possible ?
You are able to locally host n8n as well as LLMs to run your data through. To get started, check out this video: ruclips.net/video/MGaR7i35KhA/видео.html
Hi! This is odd, I approve all the memberships every few hours. I would try to request again because no memberships have been pending for a month on my end!
Whilst the agentic workflows and platforms are magnificent examples of human ingenuity, I can't help but think that we humans are already suffering from 'information overload', and that agents are simply going to make that worse as most pump out far more information than they process and reduce.
Bob: I wrote an agent to flesh out a single line symopsis to a full email! | | Alice: I wrote an agent to distill emails into a single line for easier reading!
I guess my team is not going to love this as I’m trying to replace so many people with agents 😂😂, excellent job my man. Haven’t seen the entire video yet, but do you have automations for WhatsApp as well so they can respond to small product requests and invoices ?
Hey Nate, Do you think it is possible to use the "Tools Description" as a "System prompt" instead of the main agent system prompt itself? This would allow you to easily disconnect a tool without having to change the whole system prompt, but it will just also automatically have disconnected that part of the system prompt. Do you have any idea why they wouldnt have built it like that? Let me know!😄
I think this could work for the Ultimate assistant yes, but it wouldn’t be as consistent. While testing, I was having some issues with the way data was being passed, so adding in more details and an example in the prompt helped resolve this issue. You make a good point though, but I guess the idea is once we have a function working well, why disconnect it? From there we want to keep adding more and testing to make sure they all still work properly.
I have got an issue addressing it in n8n can anyone help me to solve this issue:- "Due to Telegram API limitations, you can use just one Telegram trigger for each bot at a time" What solution i need to do??
Hey there! I’d recommend joining the free Skool community and making a post about this issue in there! Provide some screenshots and I’m sure we can help you get your issue resolved.
@@danquixote6072 I have created several bots but for each bots in N8N i can only use the trigger at once and when I use the trigger once more it's saying after that telegram app limitations issue
I agree with you. The main idea here was to show this type of agent framework and how easy it is to spin something like this up. There is so much more that can be done by stacking agents on top of each other. Inception!
![Trump's Tariff Tantrum | Benzinga’s PreMarket Playbook [LIVE] 🔴 | February 3rd, 2025](http://i.ytimg.com/vi/tH1En6djmpY/mqdefault.jpg)
![Trump's Tariff Tantrum | Benzinga’s PreMarket Playbook [LIVE] 🔴 | February 3rd, 2025](/img/tr.png)
![Noob To Max With DRAGON REWORK In Blox Fruits [FULL MOVIE]](http://i.ytimg.com/vi/LnBMOinoOvA/mqdefault.jpg)
![Blox Fruits Dragon Rework Update [Full Stream]](http://i.ytimg.com/vi/EqDAp8udhm0/mqdefault.jpg)
IMPORTANT: Import the workflows by opening up a new workflow, click the three dots in the top right, and select "Import from file". You must make sure when you are setting up the workflows as a tool for the main agent that you are choosing the correct workflow within your n8n environment.
Amazing work!
Whare can I find the workflow file to play around with
@@k.Valeroin my free Community! Go to 6:17 in this video to see how.
@@nateherkjust signed up waiting for confirmation to. Join the community
Yes that's how it should be, an agent delegating an army of subagents! Let's do more of that!
I just started building on Make, but n8n looks like it's a big boy version of Make... Very impressive build Nate. Just subbed, Membership Pending.
I really wonder how a system like this can be integrated inside a website, in the user's dashboard for example, I'd love to see something like this, or at least for you to outline the steps needed to do this.
Thanks for the great content!
Wow, powerful. Your content (including previous ones) are superb... Tbh, your content is one of the best for n8n AI
Thank you!
What? This is awesome and scary at the same time.
Great one I really like the error handling from the agent area. This really makes it even more of a challenge to not consider an end for a truly agentic framework.
Btw you showed your Tavily token in the video.
Thanks! I love the way you can have them communicate, it really makes me think about the different ways we can make workflows even more autonomous. Don’t worry, I always delete any keys after I publish a video :)
Truly epic n8n workflow, thanks for the share!
Whoaaa!!! That was an awesome walk through! Thank you for sharing your methods
Wow! Thanks for this content man. Love watching your n8n tutorial videos.
Perfect . I builded this system and its working really great
I love your channel
Thanks for this video and for the skool. This is very helpful as a primer.
Amazing! Thanks for this content!!
Hi Nate,
Thank you for your content-it’s very useful! I really appreciate the videos and the n8n flows you’ve shared. They’re awesome!
It would be great to see more common scenarios from a daily office job, maybe something involving a CRM API agent.
Keep up the great work!
Thanks for the content. Top class!
Great stuff!
lol. You completely caught me off guard with the Michael Scott @ Dundler Mifflin email. Keep up the awesome content!
I'd like to see more custom nodes/js/python/hugging face models etc. I find all the system level stuff difficult to learn, n8n is helping bridge the gap for me.
I love a good Office reference
Sweet 🧁!!
Your automation and content are impressive! If I join the group, what exclusive advantages do members gain? Is the leverage 2x, 3x, or more?
This is awesome thnx, What would be better is if this were an MS Office workflow.
Super easy to switch out to MS office
Please make a video on how you go about solving errors with the ai. For example, i tried making a meeting scheduling agent to use google Calendar and have spent hours trying to get it to stop messing up the available times. It says that completely empty days are full and sometimes ignored if some things are already scheduled, and randomly says available times. For example, on a fully empty day, it might say: were available from 9-10:15 11-3 and 3-4:30.
With AI agents, troubleshooting really comes down to 2 things: 1) repetitions and 2) trial and error.
Especially when you’re building something from scratch so many things can go wrong, but understanding why they went wrong is what helps you learn and prevent them in the future!
To me it sounds like you should check if it has the right source data, and then if the context is too long - as you get closer to the content length limit some of that context will be forgotten / hallucinated about. Format and iterating over slices of data, or utilizing RAG approaches become necessary
Where can I get the template file?
+1
In my free Skool community! I mention this in the video!
Do you use local or cloud n8n?
Both. Depends on what I’m doing, each has their pros and cons!
@nateherk Can I run telegram api on local n8n?
How can I make a local host n8n version of this?
I’ve been wanting to give commands solely to my
Laptop and every once a while connect to the internet but still connected to my local server ? Is that possible ?
You are able to locally host n8n as well as LLMs to run your data through. To get started, check out this video: ruclips.net/video/MGaR7i35KhA/видео.html
Hi! My membership to your Skool community has been pending for a month. How can I access the template, please?
Hi! This is odd, I approve all the memberships every few hours. I would try to request again because no memberships have been pending for a month on my end!
Whilst the agentic workflows and platforms are magnificent examples of human ingenuity, I can't help but think that we humans are already suffering from 'information overload', and that agents are simply going to make that worse as most pump out far more information than they process and reduce.
Then I guess it’s all about prompting them to actually make them useful
Bob: I wrote an agent to flesh out a single line symopsis to a full email!
|
|
Alice: I wrote an agent to distill emails into a single line for easier reading!
can I do AI agent it in other app or website
Yes. Your front end can be almost anything if that’s what you’re asking.
I guess my team is not going to love this as I’m trying to replace so many people with agents 😂😂, excellent job my man. Haven’t seen the entire video yet, but do you have automations for WhatsApp as well so they can respond to small product requests and invoices ?
Right haha! I don’t currently have any tutorials on that, but loves of videos about n8n and agents that should help get you on the right track!
👏🏻
Hey Nate,
Do you think it is possible to use the "Tools Description" as a "System prompt" instead of the main agent system prompt itself?
This would allow you to easily disconnect a tool without having to change the whole system prompt, but it will just also automatically have disconnected that part of the system prompt.
Do you have any idea why they wouldnt have built it like that?
Let me know!😄
I think this could work for the Ultimate assistant yes, but it wouldn’t be as consistent. While testing, I was having some issues with the way data was being passed, so adding in more details and an example in the prompt helped resolve this issue. You make a good point though, but I guess the idea is once we have a function working well, why disconnect it? From there we want to keep adding more and testing to make sure they all still work properly.
I have got an issue addressing it in n8n can anyone help me to solve this issue:- "Due to Telegram API limitations, you can use just one Telegram trigger for each bot at a time"
What solution i need to do??
Hey there! I’d recommend joining the free Skool community and making a post about this issue in there! Provide some screenshots and I’m sure we can help you get your issue resolved.
Can't you just create another telegram account (under another email)?
@@danquixote6072 I have created several bots but for each bots in N8N i can only use the trigger at once and when I use the trigger once more it's saying after that telegram app limitations issue
works rlly well but its going to be obselete once apple adopts this kind of thing
I agree with you. The main idea here was to show this type of agent framework and how easy it is to spin something like this up. There is so much more that can be done by stacking agents on top of each other. Inception!
FIRST
Indeed 🌟