n8n

Okay.

So you actually originally built this in N8n, and then you rebuilt it in Cloud.

What broke? And also, what would you tell a recruiter deciding to build something of their own instead of buying a tool?

So I looked at all the tools and there's some great tools out there, but almost all of them promised the world and fell very flat.

You know, like everyone's mind goes to these things.

But I started building in N8N because that was the way to do it.

a couple months ago, which I still can't believe that we're advancing this quickly, that N8n is now kind of obsolete and Claude Code can do everything and more.

But when I started building in N8n, so it wasn't super simple.

But another thing I also wanted to mention is I have a daily briefing that gets sent to my email every day that just gives me basically all the information stuff that's going on in the world with AI, like if Claude has a new update or maybe like there's a new AI like tool or something that's, like growing, or maybe there's a new free AI tool that does something that I paid one used to do.

And so I currently have that set up as an automation in N8N because I started off using N8N before I used cloud code.

But I am actually going to set that up to be a cloud automation, just like you have, that goes off early in the morning for me, maybe around four or five in the morning.

And the reason that is, is because...

When I first kind of reached out to my parents, I was learning AI and saying, hey, you guys own a business.

You want to check out some automations and stuff I built? And so then they kind of, that was the point where I started working more with them.

And the first automation that I worked on, I don't know if you guys remember this, mom and dad, but it was a podcast guest outreach workflow that I was building at N8N.

And so it was pretty much the same structure as this, where it would look through Apple podcasts or different podcasting apps and scrape that information, find the email addresses, and then send them emails, inviting them onto the podcast, just because we were trying to get more people on.

It is not, for the most part, something any... anyone in the profession would personally be building or touching.

And it might be, I mean, you might have your own agentic AI that you orchestrate and you've built a whole N8N workflow and integrated with everything and more power to you.

The bulk of the profession, not there.

agree agree byron brings up a good point as well about the monitoring of what's happening because i think a lot of people start to just let it go but from a process and computational standpoint someone should actually be looking at these things and we should be asking our vendors to give us that visibility into what is being asked and because if everyone's i'm going to go back to the example i gave if everyone keeps asking for sales information Because it's not available in the dashboard or reports that we have.

And that's why, in the end, we stopped doing that.

And we went other ways that... um that made more sense but things that work very well is asking asking your agent to assess how easy it is for an llm to understand your website who better to assess than your agent here's my website do you get it like can you look at companies that are very good at being found by llms and can you look at what they're doing that we are not doing look at their page layouts look at what files they may have and if we don't have them can you create them for me so that we can also put them in place this was like me just back and forth for an hour and we upgraded like dramatically our our way of of being found and understood to the point that now any ai i put it in it's like hey do you get what we do yes blah blah blah and it is list out what the possibilities are and they just know it exactly and and um that's then when you when you're at that point then creating any type of other integration let's say you want to create an mcp let's say you want to connect into make.com or zapier or whatnot it is literally a click of a button like most of these things because the ai understands your api because you've optimized for that it understands exactly what you do um like Zapier, N8n, and SoonMake as well, they all have CLIs, so they have programmatic access to create integrations.

You literally just create a repo, you tell it, here's my API, make me an integration, please.

And then you'll get some feedback from when you try to publish it, and you just feed it to your AI again.

So it used to be hard to run open source software.

It's like, yes, here's a package that you can install.

Let's take N8n.

Actually, it's a good example.

It's not a Gaia company.

It's a project by itself where if an agent wants someone to come and work on an N8N workflow, build a quick intake sheet, build a quick this and that for them.

We have resources that can go and nerd out for a couple of weeks to bring stuff to life.

In the age of AI and cloud codes, a lot of SMBs are wanting and wanting to see a full-blown business apps built just for them.

Or the Frontier Labs themselves may come out with a tool that lets you do more of this without having to code anything.

But if there were a drag and drop way like, you know, N8n version 2.0 that you could build an agent harness, that you'd have employees doing that more.

That's helpful.

We have about 10 minutes left here.

So you might use a separate tool to do the prompting or do the kind of interconnect.

Some people still use Zapier or Make, like Integrately it was previously called, maybe N8N, something like that.

Basically, you've got this trigger, which is transcription recorded, and then there's some sort of automation that follows from that.

And you could give it a reasoning prompt.

Now how do I know this? Well, at g2 we've been using cloudflare data to track traffic from userbot okay 2025 we had a million human beings come to us directly from chat directly from chat it was part of how we bent the curve on traffic why because we obsessed with G2 citations being related to our best of software awards and all our best of category content instead of just pattern matching content.

So where G2 tended to show up in our citation counts was at the bottom of every return as the proof between why they recommended AgentForce over UiPath over N8n.

And that's where we were actually able to get meaningful traffic.

So if you care about traffic, you have to care about citation position.