Thank you very much.

It was a very nice talk.

Very polished product that you already have.

Introduction

I'm not going to have such polished product, but I have something that is some kind of like in the life of an AI developer.

I'm myself an AI developer.

I work at Piango.

It's my AI agency in Switzerland.

And we built a tool that sends cold outreach emails for you in a way that makes sense and that looks like a human has written them.

The Problem: AI-Generated Outreach That Feels Like Spam

Why we do this or why we wanted to do this is because we receive a lot of AI emails every day that you can tell immediately that it was AI generated, that it's kind of spam.

You don't even look at it.

You throw it in the trash or even have a rule.

My dad has a rule that says like quick question and then he has a rule for this quick question and he moves the emails

into the trash

The Goal: Human-Like Cold Emails Backed by Real Research

so what we wanted to do is we want to build a system that makes

Emails as if you would do the cold outreach and you spend time on that on that business that you want to contact

Spend time on research and collect the information that you want and then send a carefully drafted email out

We created the system like a month ago or so

and already starting to use it now.

I'm here with Roya and Katie.

They also both work at Fiango.

2And we have sent so far 270 emails out.

And basically, click -through rate is very high.

And we got some couple of responses.

And people don't notice that it is written by AOA,

which was exactly the point that we wanted to achieve with that.

that.

Early Results and Who This Is For

So yeah, nobody has time to do this.

I talked to a friend, he has a wine business,

there is a slide later.

He basically lives off that.

He sends email as his job.

Without

it, he would not have any revenue at all.

So that's kind of the target audience for

such a system.

We are an AI company, we need customers to make software for.

We use that

tool as well.

There is basically no limit if you have to reach out to other companies

companies, for your services, for your products, whatever it is, that's kind of the thing that

would be the thing for you.

It basically makes everything 10 times faster and also 80 % cheaper because you don't have

to spend manual labor on it.

And since we are an AI company, we know how these systems work.

So we build them all the time for our clients.

So we have 10 years of experience being a software development company.

and that's why we can we can do this so effectively today with AI if you're

How the Cold Outreach System Works

Step 1: Define Tone, Signature, and Campaign Criteria

interested in the background actually my AI agents are building stuff for me and

I'm making plans for them so if you can see like this is how we are using AI at

work effectively our cold outreach system is built in like a couple of

stages you create kind of how you want to the emails to be sounding you add

add your email signature, then you create a search query

to look for a certain area for businesses.

Then you tell the AI system, for example, hey,

I want businesses in Aarau that are wine shops that

have terrible websites, for example.

This could be your campaign angle.

Then you run the pipeline, and then you review the emails

and send them out.

1So the review process is not completely automated.

It's still you look at the emails

and send them out when they make sense.

That's kind of what I'm going through here.

So you can just add how you would write the emails usually.

The longer you write them, the more AI has to understand your tone

and how you're going to sound human, basically.

Then, of course, you can add your signature.

This is just Katie here.

She added her signature to show you.

you.

Then you say, okay, I want sushi restaurants with bad websites in Zurich, for example.

And then you pick an angle.

They may have also a poor mobile experience, whatever.

And

then you hit run and you can see there is emails created for you.

You can approve them

and send them out and wait for the response.

That's kind of the end result that you get.

Step 2: Automated Research and Email Drafting in the Background

And what is actually running in the background is that Google Maps scrapes the leads from a certain area.

Then it does an email discovery.

It uses Hunter .io, but it basically doesn't matter what kind of service that you use to get these emails validated.

We use Hunter .io and Whois, whatever other way that there is that you want.

Also, of course, the website.

It goes through the website, it does a lighthouse check, an SEO crawl,

that's all for our system.

And then it uses cloud to draft the email with all that information.

So

it takes everything into account that it did.

And it uses an email provider, an email service to

send it out so you can get the email stats afterwards.

And yeah, this is basically how

Adapting the Pipeline to Other Outreach Scenarios

we did it before, but if you want to do it, it doesn't have to be exactly this pipeline.

It can be adapted to any use case, any scenario that you want.

If you, for example, don't need a lighthouse check or an SEO check for the websites, but

you just want to send out emails for your product, for Schooly for example, if you want

to reach big, large corporations, that could be a campaign that you can set up, then you

don't need these kind of technical tools, but you might want to extend the research,

research, the AI research significantly, to draft a nice cold outreach email.

For example, as I mentioned before, this was for the wine shop.

He needs a significantly longer research process and less technical analysis to make sure that

his outreach email is crafted in the way that he wants, like with emotions and experiences

experiences that are more in line with his business.

And we do all this with AI in basically record time.

Speed of Execution: From Idea to Deployed System

So we set this whole system up in like three hours,

and then it's up and running and deployed and can be used.

And I have a demo for you that I can show you.

It's not going to be pretty, but it's usable for our use case.

Why Django (and Why AI Writes It Well)

So what you see here is the Django admin interface.

And a quick note,

Piango is actually the combination of Python and Django.

It's a very old web framework, 15 years old.

And that's why AI is today also so good

at writing Django code because it's so old,

has so much content on the internet,

so many stack overflow questions.

So it writes basically perfect Django code

in the first goal almost.

post.

And that's why we chose to use this.

You have such low friction to go to production.

Demo Walkthrough: Analytics, Campaigns, and Approval Flow

So what I can show you, the email analytics.

This is a page that just shows you what the

deliverability is and if people are looking at your emails, if they click on it, etc.

And of course, you can see all the emails that were sent out.

I can go to the, we call

them scraping orders, but these are basically the campaigns.

When we look at one of the

campaigns that Roya made today, for example.

We have dentists in

treats.

And she just added a search query to then created a

campaign goal.

This is I think she did this manually to make

sure that it's a proper campaign.

And then it's found 16

leads in a certain area.

If we scroll down here, we can see the

exact area where that is, and what the leads are.

And from

here you can go to the lead of course or you can just check a list of all the

leads here and from here also then look at the emails and approve them right

away so you can see this is about like the cold outreach email that was

generated basically of that information and we also added a feature so she can

see it in in English so because she doesn't speak very well German yet but

wants to send emails in German as a cold outreach email.

She can check it and then

approve it from there.

There's a lot of other things that I can show you here.

The most important thing is just this list of campaigns, how much the

deliverability is, when it was created, who is managing it, etc.

And then

Follow-Ups, Replies, and Lead Database Views

there's of course follow -up sequences that you can look at.

You can see all the

the email replies that came in.

Of course, you can have an overview of all of the leads.

That's just basically your database.

And from here, you have more information

of what happened with that lead.

Was it scraped?

Is there an email address?

Is the quality high?

Is there a high rating or a low rating on Google Maps, et

cetera?

Wrap-Up

Yeah, and that's kind of already the demo.

There is not much here to see.

It's basically a tool that we created

that we use once a day to make a couple of campaigns and then throughout the day

we approve the emails and this way we reach a much higher amount of people in

a very short time.