Free YouTube Transcribe

Video transcript

I Let an AI Agent Build Lead Magnets for Me (Here's What Happened)

Max Mitcham · 1,240 words · 6 min read

Want to search this transcript, jump the video from any line, or download it as TXT, SRT, or VTT?

Open in the transcript tool

Full transcript

0:00Strap yourselves in, guys. This is

0:02probably one of the most insanely crazy

0:05agents that I have built. And we're

0:07using three tools for this. Triggery,

0:10Clay, and NA10. Now, what if you could

0:14create personalized lead magnets at

0:18scale? And so, what I mean by this is

0:20let's say you reached out to 100 people.

0:22Imagine if you could create a 100

0:24different lead magnets. that changes

0:27based on the job title, the company, and

0:30the type of content that that individual

0:33is engaging with.

0:36So stay with me and we're going to dive

0:38into this one. So first things first,

0:41what we're doing is we're using triggery

0:43to monitor anyone engaging with a

0:45particular piece of content. Okay. So

0:48here I am sending through leads of

0:50people engaging with particular thought

0:52leaders where they're engaging with

0:55particular content within that thought

0:56leader and it goes through to Clay via

0:59an evergreen campaign which in layman's

1:01terms means it never stops running.

1:04Anytime this particular thought leader

1:06that I'm monitoring posts, I then track

1:08that data and send it through to Clay.

1:12Well, you can start to see it coming

1:13through here. We get so much bits of

1:16data with uh this particular table and

1:19we don't use any credits which is cool.

1:22So first things that we're doing is

1:23we're just kind of fleshing all of this

1:24out and then we start to do uh an ICP

1:28check. Are they based in the location

1:29that I want them to be? Yes or no?

1:34Then is the post that they're engaging

1:36with containing data or does the post

1:40contain information around social

1:43signals, ABM strategies, sales

1:45automation, personal branding? So things

1:47that are relevant to me from triggery.

1:50If they are, then I want you to go and

1:52run some research on why this individual

1:56would would be engaging with that type

1:58of content based on what that post is

2:00about. and then understand if we could

2:03build on that post that they've already

2:05liked, how could we write a kind of

2:08research paper or a lead magnet based on

2:11that? So, for example, this one was

2:13like, how can B2B sales consulting firms

2:15help organizations overcome ABM platform

2:18adoption challenges to ensure successful

2:20implementation and value realization?

2:23And this kind of changes, right? Like

2:25depending on the type of post they're

2:27doing. Here's another one. How can AI

2:29powered personalization in B2B outreach

2:32be leveraged to optimize website visitor

2:34conversion rates and sales efficiency?

2:37So what happens is this basically

2:40triggers um what we call the HTTP API

2:44here which basically just sends this

2:47data straight through to this particular

2:50NA10 uh workflow via a web hook. Now

2:54what then happens is a sequence of

2:57events and we're using 1 2 3 4 five

3:00agents inside of this particular flow.

3:05So first of all that particular subject

3:08so if we go back to this one how can AI

3:10powered uh personalization power B2B

3:12blah blah blah gets sent through to this

3:14which is the query builder and I've

3:16already ran one and this one was on how

3:18you could use trigger fire to book 30

3:20meetings. So we'll go with that for the

3:21example. But the idea is that that

3:24information that Clay has just found

3:25gets sent through to this which is the

3:27topic. From there, this agent comes up

3:31with four different um conversations or

3:34kind of research questions that it needs

3:36to do. And that's all that this agent

3:39does. It just researches this particular

3:41topic and breaks it down into four

3:43subtopics.

3:45From there, it then sends it through to

3:47what we call the research leader. the

3:50research leader will create the initial

3:52research um for that particular

3:56individual or those particular subjects.

3:59And so how this then basically looks is

4:01it takes those four topic areas which is

4:03the ones that we looked at previously

4:05and then starts to run some deep

4:06research on those topic areas. From

4:10there, it goes through to what we call a

4:12project planner, which then starts to

4:14create the title, the subtitle, the

4:16introduction, the different chapters

4:19that are going to be involved in this

4:21research paper slash um lead magnet. And

4:25so, think of it like this. The project

4:26planner takes all that research that was

4:28just previously done and then works out

4:30what are the chapters that are needed

4:32and then it works out the prompts that

4:34are needed for that chapter. Because

4:38then it sends let's say those five

4:39chapters through to this team of

4:41research assistants which then takes the

4:45prompt and the chapter title and has all

4:48of the context about the article itself

4:50and then starts to run deep research on

4:54that chapter. And one of the key things

4:56that we're doing here is we are

5:00basically giving it the context which is

5:03hey we are writing this chapter the

5:05previous chapter was on XY Z the next

5:08chapter is on XY Z here is the prompt

5:11for you for this particular chapter and

5:14then it runs that research and then

5:15compiles it for you where it then sends

5:19it through to this next uh editor here

5:21which is the agent which just kind of

5:23combines it all together. It kind of

5:25checks grammar. It checks the style of

5:27writing. It makes sure that it copies

5:30that particular flow and so much more.

5:33And it also writes it in the language of

5:36the prospect themsself. So we know based

5:39on the location and the content and the

5:43post language that the individual is

5:45engaging with what language they speak

5:47and then the magnet is then written for

5:49that individual in that language should

5:52you wish to have that kind of option

5:53turned on.

5:55Some key things that we're doing here is

5:57we're using Triggery's knowledge base.

5:59So I have a trained model that is

6:02basically trained on everything that I

6:03do. So it uses that to help answer any

6:05triggery based questions if I'm wanting

6:07to use Triggery in the lead magnet. And

6:10we're also using Perplexity's deep

6:12research here.

6:14Once the editor has done its work and

6:17its magic, it creates a Google document.

6:20It adds that text to the document. It

6:23then makes this document sharable and

6:25then gets me the link where it then

6:27sends this data back to Clay.

6:31Um, let me pull up the right one

6:34with the documentation. So, here's the

6:36share link, which is really cool. And

6:39here are the individuals coming through

6:40that it needs to then send back to. And

6:42then you basically can just reach out to

6:44those individuals with that share link.

6:47And so you can say like, "Hey Christian,

6:49here is the um here's like the lead

6:52magnet that I've compiled or whatever

6:53kind of copy that you'd want to do." But

6:56this then works on clockwork for you.

6:59Anytime someone says yes, they would

7:01like to see more information. Or you can

7:02just chuck this in the very first

7:03message if you're reaching out via

7:05LinkedIn, whatever. And it is 100%

7:08personalized to that individual. Now, if

7:11you want this framework, let me know

7:12below. I will share it with you. the

7:14clay table, a full video breakdown on

7:18YouTube as well on how to run this thing

7:20and the NA10 template and how to set up

7:23trigger fire in order to do this. Just

7:26let me know.

This transcript was generated from the captions YouTube publishes for this video. Get the transcript of any YouTube video atfreeyoutubetranscribe.com: free, unlimited, no sign-up.