AI Personal Assistant 2.0 | This Agent Calls Other Agents (No Code) in n8n

[@rafss99]

Perfect timing. Calling multiple agents and using fromAI was what I needed. Keep up with the great work! Your videos are on point, no fluff, super helpful.

[@NataliyaAi]

Thanks, Nate! You've got a new subsriber ;) Really impressive stuff!

[@VisibleProjects]

Thanks again - as always Nate. Your videos are guaranteed to add another nugget to my war chest of knowledge.

[@mateussaviolo]

I'm Brazilian and I'm learning English to study your videos and I tell you: If you launch a course from basic to advanced, my money is yours. 😄

[@Gesssy]

Isso aí. Tb estou treinando inglês com esses vídeos

[@Pedro_forlife]

Tmjjjjj eu tambem

[@sandowssecret7363]

this is next level bro this makes things truly modular thanks

[@simongentry]

hey N. great idea. so much possibility. major research possibilities... news, social media api's and scraping!

[@OmariBroussardHSCEO]

This is what I’ve been looking for!

[@vivipessoaUGC]

You are so good!!! Thank you for your videos!!!❤🎉

[@GeringSa]

Amazing. Very good videos and tipps....keep it coming!

[@nateherk]

Thank you sir!

[@frankdearr2772]

Great topic, thanks 👍

[@Mjefferson001]

nice work clear concise ! stay crispy

[@fesouto5381]

Amazing content Nate! Thank you and congratulations, every day you outdo yourself 👏🏽👏🏽 I’m making something similar, and thit video helped a lot 🙏🏽

[@ZukunftBilden]

Please make it so it doesn’t only seem relevant to people who watched and liked your last video. I want you to hit 10k. Make a title and thumbnail that is not a 2.0 - an improvement. But more like - the ultimate expandable multi agent super agent. Or something in that direction.

[@nateherk]

Thank you for the feedback, great point

[@loganphillips7749]

Would love to see an AI Agent that can fully operate Todoist. I'm a little stuck on this. Todoist isnt available as a tool, but is available as a node.

[@simfor]

❤ thank you

[@Sam_Bent]

OVER 1k likes, when is the walkthrough coming?

[@nateherk]

It’s been out for a while man! Check the channel for the step-by-step.

[@nateherk]

Link in the description of this video

[@Sam_Bent]

@@nateherk Thank you!

[@SimeonPreu]

"steal this" as mentioned in the thumbnail would be much easier if the JSON for this project was somewhere to be found...!

[@nateherk]

All of my workflows are available for download in my free Skool community

[@Gismotronics]

I'm wondering if it could securely/safely get bank and other credit/debit card transaction/balance info? Would be great for managing finances. For example, if you trying to your spending withing a monthly budge, you could setup some checks and simple reports to let you if you're on track.

[@harveyroberts2274]

Im new to N8n, and thank you for the videos. Keep up the good work. Just a quick question. In this example, would this count as 5 workflows or not.

[@nateherk]

This should only counts as one active workflow (the parent workflow)

[@tomthompson1198]

i want somethign locally ran and or hosted. look forward to that.

[@theshnizter]

That's good only if the end user text one msg at a time.. But normally ppl text multiple messages at one time which confuse the AI response

[@nateherk]

The AI can still handle this. Really comes down to fine tuning the prompting at that point.

[@germanomatta]

Nice Video!

[@davidgriffey5045]

This is great! I have it all built but am having problems. I switched out the openai parts with ollama which was fine. BUT even tho it can read my google sheets for the contacts it wont add to it. it says it does but it actually doesn't add anything to them. even made a todo list the same way and nothing it wont edit that either

[@nateherk]

I’m sorry to hear this. I’m sure other people have had a similar issue, check out my free community and see if anyone in there can help you out! Shoot me a DM as well!

[@ctcopeland44]

How do you attach the 'Success' and 'Error' fields so that the agent knows which response to return to the main agent? Thanks for the awesome vid!

[@stephena8965]

Go to settings and select On Error and select "Continue (using error output)"

[@SamFouladgarSuccessCoach]

Thanks a million Nate! BTW how are you getting the node tools such as google calendar or gmail directly. I can't do that. Is there anything I need to do to access those nodes?

[@funnydogfargo1026]

You need to update your n8n.

[@nateherk]

Thank you for the kind words! Make sure you’re on the correct version of n8n and looking for tools to call from an AI Agent.

[@GuruSanketBiswal]

How much does it cost? Is their costs involved in multiple integration individually to get access to their respective apis?

[@CrispinCourtenay]

Curious if there is an upper limit on n8n for max # of agents before you start to hit either the performance wall or start to throw errors?

[@Nuetzt-ja-nix]

Great video - helpful as always. 1) Is there a way to build something which is iterating? Like: I would like to meet with Dan for dinner. I can do Friday, Monday and Tuesday. Ask Dan which day would be best and then pin it into my calendar and inform him. 2) Is it possible to build something to first check availability in your own calendar before creating a new event. I mean iy you accidently ask you agent to book a second dinner at the same time, that would result in double booking. Not so?

[@nateherk]

Thanks for the kind words! You make 2 great points. 1) This is possible yes, it just adds a few more steps into the agent and passes along more variables. 2) You’re correct, I’ve made agents that first have to check availability to make sure no double booking occurs. Wanted to keep this build simple enough to run through in under 30 minutes, but I really appreciate the questions as that kind of stuff is necessary for this assistant to be fully practical.

[@stevenbaert1974]

Great videos! Anywhere we can find/download the n8n flows of them?

[@nateherk]

Yes! In my free Skool community I post all of my workflows. Link in the description, see you in there!

[@irajatsolanki]

Can we add calling support and reminders through calls

[@nateherk]

That would be a great way to improve this agent! You’ve given me some ideas…

[@tomthompson1198]

where are the prices, token and other performance metrics?

[@Businessology.]

Great vid! Do you know if it's easy to connect a n&n workflow with an front-end on Cloudflare?

[@nateherk]

Hey! Yea it should be relatively straightforward. I imaging you’d expose your n8n workflow as a webhook and deploy the front end so they are able to talk to each other.

[@RIyaadhMahomed]

Docker in

[@mishaeldada9766]

Hey nate do you think we can replicate the sequential approach where chatbot has predefined path/questions for users to answer just like we have on tools like Voiceflow and Botpress on the n8n chat interface too?

[@nateherk]

So you’re looking for a sort of specific script for the n8n agent to follow? This could be worked into the prompting for sure.

[@jaspervanharten6124]

How much do u pay for the input and output tokens from openAI? What does it cost to run this bot?

[@nateherk]

Hard to give an exact number, but it is fairly inexpensive. For example, during the time it took me to build and test the agent, I only used about $1.34 with GPT4o

[@petarvukovic1181]

Hi Nate how to sell this and similiar autoamtions on market?

[@nateherk]

There are lots of ways to monetize these solutions. If you'd like to dive deeper into this stuff check out my Skool communities!

[@cerebrumai]

n8n version 1.72.1 (Self Hosted) Time 1/1/2025, 4:33:56 AM Error cause { "level": "error", "tags": {}, "extra": { "workflowId": "0NtlJ41IozGhtFa6" } } but the agent is there ?

[@Teamsao90210]

Am i right in thinking it cant edit events , so if the client suggested a different date and time it doesnt as is in the video, have the ability to change the date, time, location etc?

[@nateherk]

You 100% could. It would just involve tools that can grab an ID for the event in order to update/delete it.

[@LePsyclone]

Did you manage to get n8n to work with Microsoft teams calendars ?

[@nateherk]

I haven't done much integration with teams at the moment. Does it seem to be more difficult?

[@meowhib]

Can we use the whisper model so we can send the bot voice messages?

[@nateherk]

100% - it would transcribe the audio and then feed the agent that message! The logic would all be the same after the initial request

[@Petori_Hustler]

is it possible to build one without credits...

[@nateherk]

You can look into self hosting and running your AI locally

[@damz-lare6616]

can you do one with whatsapp

[@akiempaul1117]

I think you need to work on your SEO, I tried googling for doing this with n8n and you didn't show up

[@nateherk]

Great point, thanks for the heads up

[@JakeRichards]

make the tutorial! gotta see

[@nateherk]

The tutorial is out! Check it out: kzbin.info/www/bejne/kIGXl52MjLCnaNk

[@JakeRichards]

@ dope :)!!

[@zachisparanoid]

Somehow, i could tell this is a Tom Misch song...

[@nateherk]

Haha!

[@andrewl5095]

will this work on whatspp?

[@Pedro_forlife]

Hey Nate is it possible to build this same Agent on WhatsApp ?

[@nateherk]

Hey Pedro! Do you just mean the way you talk to the agent and how it responds? If so yes of course. We could also build an agent to take actions within WhatsApp if that’s what you’re looking for.

[@PhilipDenys]

1000likes

[@gu9838]

neat but once again not free and it costs per amount of useages soooo

[@PixelFrontier-channel]

It's disturbing how trusting you guys are with open AI and your personal data.

[@mjsavage711]

If your concerned you can run it with ollama and open source models, I use Llama 3.x models a bunch and even embedded my vector stores with open source embedding models. You don't have to use just one provider, what makes this automation so powerful.

[@nateherk]

For a client project, would always use a self hosted version of n8n

[@SiraEkabutPlus]

don't worry... your personal data is already al google or microsoft already 😂

[@edoardododoguzzi]

The problem is the personal data is not personal from the first day of your birth...

[@hevra5991]

@nateherk 🎉🎉 Great video! You should consider using your own model from Ollama, running on your own cluster and server with its own database. This would allow you to keep your sensitive information separate from OpenAI and other models.

[@madisopabul]

this is shit i hope this message find you well i would never read it

[@nateherk]

Hmm

[@madisopabul]

@@nateherk hmmmmmmmmmmmmm