AI Automations as a Service

I always hear tropes about how AI is going to replace 50% of the workforce, and I have been interested in learning about automation and using AI tools in order to create a productized service offering…

So I’m documenting my process of learning.

Specifically, this project will be about keeping a journal of what I’ve learned and how I’m progressing toward the goal of getting a business to pay me for building them automations that either save them time, money, perhaps both.

This should be interesting because I have no prior understanding or experience in this.

So today, I’m considering going thru N8N and MAKE’s free resources, as well as beginner course I came across.

What I’ve Learned Today:
-To sell an automation, it must save the business time or money (or both)
-It must produce clear deliverables (must be a quantifiable deliverable that adds value). It must be able to answer for; “what did this automation do for me over the last month?”
-Templated automations minimize things going wrong, which make templated automations a greta starting point for beginner automation engineers to sell.
-One trending automation people are selling is a CRM for creative agencies to more efficiently manage assets.
-It’s likely better to start by selling an automation that works well before I try building my own automations.
-A lot about scraping
-How to get clients
-Which automations to sell
-MAKE basics

Scraping:
So basically, there are two types of websites; static and dynamic… Static websites mean there is a file on a hard disk on that server with all the website’s HTMML and it doesn’t change. Static sites are super easy to scrape with MAKE or N8N, where dynamic websites require you use a 3rd party platform. Many of these 3rd party tools integrate with MAKE or N8N so you just need to add an extra module. Dynamic websites on the other hand are a set of rules in a different programming language where requests to the server are applied and HTML is generated and sent back. Note that there is contention between whether server-generated vs. static-generated websites are quicker/better.

When you scrape a website you get HTML; but HTML is not what you want… What you want is some type of structured data. That’s where we need to understand parsing. HTML is for machines, and parsing renders that info for people. Parsing is simply breaking down the HTML into constituents and extracting the bits you want.

Getting Clients:
Scrape leads for outreach > Record custom loom videos > deliver videos to leads. Can use Apify to scrape. Aim to scrape 4k profiles which should amount to 2-3k emails.

When it comes to getting clients there are 3 acquisition channels:

1. Cold Email

   • 9 mailboxes + Instantly = $187 per month

   • scrape 3k emails via leads rapidly = $9

   • 25-30 emails per day per mailbox

2. Upwork

   • send 5 custom applications per day = $2 per application because Upwork is pay to play. This is recording a video of you solving the problem for them because that’s how you stand out when it’s pay to play.

   • bright yellow photo background in profile pic to stand out

   • turn “online for hiring messages” on

3. Communities

   • join free/low-cost communities targeting your market

   • read community posts > jot down problems > post the solutions

   • create 20 posts per month

Which automations to sell:
-Deep Personalization System: Apify is basically a marketplace where you can scrape stuff. They have an Apollo scraper, which means you can scrape their lead aggregator.

-Search Intent Lead Scraper: This is scraping job listings instead of people. If a company wants to hire a [sales rep], then we know that company is willing to spend money on [getting sales].

-Youtube Parasite + Content Improvement System: This scrapes youtube videos, gets transcripts, and uses AI to find gaps in the content to improve. AI then writes better content so that you can repurpose it for social media, youtube scripts, etc. It’s best to use a 2 pass system for social content AKA where a human is looped in so after it’s finished creating something, you add it to a queue and you have a human tweak it.

-Automated Newsletter Production System: This finds trending posts on reddit for a specified niche and filters them suing AI and then curates those posts into a newsletter-style content format.

MAKE Basics:
MAKE is a low-code automation platform that is generally recommended for beginners like myself getting acquainted with building automations for the first time. The Pro plan is currently less than $20 per month and is advised for beginners.

All automations in MAKE start with a trigger. A trigger is what triggers or starts a flow. For example, “Watch New Row” is a common trigger that starts a process within a google sheet:

Questions I had while learning and the answers to them:

• What is a webhook and what is it used for when building automations?

  • It’s how apps send real-time data to other apps whenever something specific happens.

  • In the context of automations, it’s like a doorbell - when someone hits it (a trigger event), a signal (data) gets sent instantly to another app to do something (eg, start your automation).

  • For example: A new order is placed on your Shopify store ➔ A webhook sends order data to MAKE ➔ MAKE updates a Google Sheet.

• Why would someone want to update a google sheet if a shopify order is placed?

  • For reporting, inventory analysis, backup records, etc. While this done in Shopify, some people prefer more customization in their analysis’

• What is JSON and what is it used for when building automations?

  • JavaScript Object Notation is a simple way to organize and store data as text.

  • In the context of automations, JSON is like the language that apps use to send data to each other. When you connect apps in platforms like MAKE, the data being passed around is often in JSON format.

  • For example: a webhook gets triggered by a new shopify order: {

    "order_id": 12345,

    "customer_name": "Jane Doe",

    "total": 49.99

    } … and MAKE reads this JSON to update a Google Sheet, send an email, etc.

• What is an API and what is it used for when building automations?

  • An Application Programming Interface is like a menu at a restaurant - it shows what you can order (or request) and how to ask for it, but it doesn't show the chef (backend) making it.

  • In the context of automations, APIs let different apps talk to each other and exchange data. If an app has an API, you can use it to automate tasks or get data.

  • For example: using the Shopify API to get orders, update products, etc.

• It seems like API and Webhooks were described as the exact same things. What gives?

  • API: You ask for data when you want it. (Pull)

  • Webhook: Data is sent to you automatically when something happens. (Push)

• What is the point of using project management platforms such as Airtable or Clickup when building automations?

  • The answer i searched for just made me more confused. I don’t know what to put here.

• What is schema and what is it used for when building automations?

  • Schema is like a blueprint or structure that defines how data is organized and what it should look like. It specifies what fields (data points) are required, their types (Eg, text, number, date), and how they relate to each other.

  • In the context of automations, schema ensures data is correctly structured before it’s sent or processed, helps your automation know what information to extract from a tool or database, and prevents errors like missing fields or mismatched data types.

  • For example: When you use a form to collect data, the schema might define: Name (text), Email (email), Date of Birth (date)… When your automation connects to an API, it uses the schema to send data in the correct format.