AI agents for marketing are here and

it's going to totally change how you're

going to grow your business. These

basically function as employees that you

can delegate tasks to and they can just

go and run the operations for a specific

marketing tactic that you're doing. For

example, like your Facebook ads or your

Google ads or your cold outbound. And

what's fundamentally different than AI

chats from before is that these systems

can evolve. They can learn from what you

teach it and they can ingest skills that

you provide to it. So you can almost

upload knowledge into them like the

Matrix. I'm Cody Schneider. I'm the

co-founder of a company called Graft.

I've been building online businesses for

the last 15 years of my life. And in

this video, I'm going to show you how to

build your first AI agent for marketing

and then I'm going to cast the vision of

where this is going and how this is

going to entirely change how companies

grow themselves in 2026. For the first

time ever, you can have these virtual

employees that are connected to your

business data that are making decisions

about your go-to-market motions based on

what actually drives the revenue for

your company. With that said, let's get

started. All right, so for this build

today, I'm going to be doing it entirely

on Graft. It has access to my live

business data and it also has the

ability to build these agents out and

run them in persistence in the

background. So, what we're going to do

today is create an agent that does

keyword research based off of our live

Google Search Console data, our Ahrefs

API that we've given it. We're going to

have it go and research keywords, then

find what is ranking on page one of

Google for those keywords, extract the

content and put that into the context

window and then write blog posts based

off of what is currently ranking and

publish those to the blog and then turn

this into a recurring action that it's

doing on a daily basis. All right, so

let's get started and actually do this.

So, I'm just going to prompt it to begin

with something like this. All right, use

the Google Search Console for graft.com

and use the Ahrefs API key to go do

keyword research for keywords that are

related to AI for marketing. And now I'm

just going to go and I'm going to hit

send message and the agent's going to

start working on this. And the reason

I'm starting with just the small

bite-size task is I'm going to teach it

this process, how to actually do this.

Just like I would teach an employee,

step-by-step kind of walking them

through it and then building and then I

can build this process, what I teach it

into a skill that it can use in the

future. I'm basically building the

memory up on this activity that it's

going to be doing for me. And it's just

working on the task that I provided it

still, but I just realized that we left

on the Snoop Dogg function. So, we told

the agent to respond as Snoop Dogg. So,

it's pretty funny to show this in the

demo. All right, so it just completed

that initial research, but I don't want

it to write about anything that's

already published. So, I'm going to tell

it that. I'm going to say, "Add this to

your memory so that we don't publish

content that's already been written

about. So, go and look at our CMS,

Strapi, using the Strapi API to make

sure that we haven't written about that

keyword, that target keyword, already in

the past." And I'm also going to tell it

now that we're going to be writing blog

post content. So, look for keywords that

have a search intent that a blog post

would be a better result and then let's

rank stack these keywords. So, let me

say that now. So, for these keywords or

for this function that you're going to

be doing, we want to look for keywords

that have a search intent that is a blog

post, not like a landing page, etc.

Let's go and rank stack these keywords

now and which have the largest

opportunities. I want you to rank stack

30 of them that have the biggest

opportunities for us and that are the

closest to our product. Like the search

intent is related to our product,

graft.com. And then I'm going to hit

send message, let it work on that. All

right, so it's responded back with this

rank stack list of keywords now, but I

want it to focus on stuff that's have a

lower keyword difficulty just so that

it's more possible for us to be able to

rank for. So, I can tell it that. But

just to keep the But just to keep this

But just keep the video rolling, what

I'm going to do is have it now go and

write actually one of these blog posts

and teach it that process of the blog

post writing. So, I'm going to say, "All

right, we're going to write a blog post

for AI marketing apps, that target

keyword. To write this blog post, you're

going to use the surfer.dev API key to

find what's ranking on page one of

Google currently for that target keyword

phrase and then you're going to use the

Exa AI API to go and extract the content

from those pages that are ranking on

page one. And the reason for doing all

of this is we're basically writing for

that target keyword based off of what's

ranking currently." And then I'm going

to hit send message. Now, what I've just

done here is I've now taught it how to

do keyword research and I'm now teaching

it how to actually research the content

that it should write for this article.

And I could take this even a step

further and this is actually what you

should do is you should then provide

your own views on the space, on the

industry that you're writing about so

that the agent will write both from what

is ranking currently and your own

personal perspective. How I do this is I

actually provided a transcript of my

view on what is happening within

whatever it is I'm writing about and

provide that as source material so that

when it's writing for that target

keyword, it's bringing in both of these

things. So now, it's pulled in those

search results via Surfer. It's now

going to use Exa to go and extract the

content from those search results and

then it's going to put that into context

and write the blog post based off of

that context that I provided. All right,

so it just got done writing this. The

blog post has been published via the

Strapi API to the CMS. You can actually

go to that page. This is great, awesome,

cool. We just did some work entirely by

agents, but now I want to turn this into

a recurring task for the agent. So, I'm

just going to tell it that. Now, I want

you to turn this into a recurring task.

You're going to publish a blog post like

this on a daily cadence. So, you're

going to do this whole thing, do the

keyword research, decide which keyword

is the best to focus on, research what

the article should include and then

actually publish that to the live site

and do that daily. Just one article a

day. And then I'm going to hit send

message and that's it. This is now a

workflow that's running on a daily cron

job. You just built your first AI agent

employee that's doing some type of

marketing task for you. And what I've

just shown you is just a single thing

that you can do. Any process that you

have, you can go and teach this agent to

go and do that for you. You can have it

manage your Facebook ads automatically

so that if they turn off ads that get to

a CPMR that is too high. You can have it

run your Google search ads by providing

a budget and then saying optimize for

this specific user conversion action.

You can have it manage your social

media, research posts, schedule them and

then analyze the data to see what's

getting the most engagement and

traction. You can have it do cold

outbound by finding target accounts,

going extracting their emails using the

Apollo API, validating those emails,

adding them to Instantly and then

actually responding to and managing the

entire cold outreach process. And the

biggest thing about this is that it's

connected to your live business data.

So, the agent is making decisions based

on the actions that are actually

creating revenue for your business. For

example, this blog post that we just

published has this call to action. I can

tell the agent to optimize for the

conversion event, which is the sign up

and then let that influence its

decision-making on what it writes next.

And this is all possible because you're

streaming in all of your live business

data from Graft and the agent has access

to that. This enables it to do data

analytics and you're not going to run

into any errors where you're hitting API

rate limits or having MCP failures or

truncation issues or context window

bloat on the data that you're pulling

from these data sources because this is

a data pipeline, a data warehouse and

the underlying infrastructure that

enables you to actually build these

agents. All of that is tied together.

And this is where agents have failed

largely previously is that they don't

have this accurate data. They can go and

do things, but the decisions that they

were making were bad decisions. So, the

data is a piece of this and then also

the constant learning memory is another

piece of this, which we built into these

agents. And what you just watched is

going to go live to all Graft users. If

you want to do get started with this, we

want to do this with you. Use the form

below to reach out or you can reach out

to me directly. We'll get you

on-boarded. I will personally set this

up for you. I will set up your initial

agents for your company. And if you like

this video, send it to somebody and also

follow me. I'm going to be doing content

like this on a daily basis showing you

how to build agents that do your

go-to-market. All right, later.