AI Automations 101 Guide
AI Automations 101 Guide
AI Automations 101 Guide
A PRIMER ON
AI + AUTOMATIONS
Why Automate with AI?
ChatGPT clearly has taken the world by storm. It's made us "10x faster", feel smarter, given us the
ultimate brainstorming buddy, and more. But the productivity gains don't stop there.
Smart teams are going one step further and integrating ChatGPT natively into their processes using
automation tools. Rather than stopping at copy/pasting prompts, they are breaking down complex
processes into smaller steps and delegating out that work to ChatGPT.
In this guide we'll walk you through the setup, basics, how to make a plan, and common AI
automation tasks to get you started.
The Basics
Automation tools can take a process and run it automatically, which can lead to huge productivity
gains for teams and individuals. These tools allow you to connect the tools you already use and set
up automation processes, without requiring technical expertise. Zapier is one of the most popular
automation tools, and one we use for the guidance in the rest of this ebook.
AI, and specifically ChatGPT, now easily integrate into automation tools. This lets you send prompts
to ChatGPT from directly within your automations. Think of this as being able to take a ChatGPT
conversation thread that you've designed to complete a task, and run it on repeat.
Before going deep on AI in automations, here are the key concepts to know:
To use automation tools effectively, it's important to understand the six key concepts that underlie
them. We'll break them down for you.
1. Workflows: Any automation is essentially just a workflow or a sequence of steps to complete a
task, that will happen automatically once triggered.
2. Triggers: In automation, a trigger is what gets the process started. Different automation tools
offer different triggers, so it's essential to select the right tool for the job.
3. Inputs: Inputs are the data that automation needs to work. They usually come from the initial
trigger, but we can also get additional inputs from later actions in the workflow.
4. Logic: Logic refers to the rules that determine what happens within an automation workflow. It's
created by setting up if/then statements, which can include filters, constraints, queries, and AI!
5. Actions: Actions are the specific steps that the automation takes once it detects the trigger.
These steps are specific to the tool you're using and can range from sending an email to
updating a record in a database.
6. Output: In automation, the output is the ultimate result. It's usually created at the end of the
automation, and can include things like the email that's sent out or the record that's updated.
Now that you understand the basics, we'll guide you through a step-by-step process to help you
break down your project objective and create a robust plan to use generative AI and automation tools
to achieve it.
It's important to know where your data is coming from and where it needs to go. Your data can move
between various sources and destinations.
Data inputs can range from Google Forms, to email, Shopify websites, task management tools like
Clickup, and thousands of other apps. Likewise with data outputs. The best way to start
understanding what's possible is to see whether the apps you use on a regular basis are available as
data sources and destinations.
To get a better understanding of what's possible, you can check if the apps you use are available as
data sources and destinations. Check out Zapier's app listing.
The automation trigger is a crucial component of your AI automation project. It's the event or
condition that initiates your automation process.
First, you need to examine your current workflow and identifying the moments when automation
should begin. Consider user actions, time-based events, or changes in data that would naturally lead
to the start of your automation process.
Select one specific event or condition as your automation trigger. Keep in mind that you can only
have one trigger per automation. Ensure the chosen trigger effectively initiates the automation at the
right time and under the appropriate circumstances.
Once you have your task list, we think it's helpful to understand what type of task each task is.
Traditionally, most automation tasks are simply taking data from one place and putting it in another.
You might also reformat or add logic in the process. These are examples of basic logic. However you
can unlock more complex workflows by using AI in your logic, which we'll call AI logic.
Basic Logic
These tasks involve simple data manipulation or basic operations that don't require AI-powered
solutions. Some examples of basic logic include:
Data import/export: Transferring data between different platforms or systems, like moving data
from a spreadsheet to a CRM.
Data formatting: Converting data into a standardized format, such as changing date formats or
merging first and last names.
Conditional logic: Applying if-then-else rules to control the flow of your automation based on
certain conditions, like sending different emails based on user preferences.
AI Logic
AI logic involves tasks that require AI-powered solutions, in our case leveraging generative AI.
ChatGPT is the most popular tool used for this today. By using ChatGPT in your logic, you can add in
more advanced pattern recognition and language understanding.
- Summarization
- Structured generation
- Freeform generation
- Extraction
- Classification
- Translation
- Recommendations
- Sentiment Analysis
You will also need to make sure that the inputs needed for one task are fully addressed by previous
tasks. This process may require iterative adjustments to your design. For example, one step in your
process might require searching for a specific record, but another step is needed to retrieve the
details of that specific record for use in another task.
Before implementing the plan, test each task individually to confirm that each step in your
automation process can handle the inputs and outputs required. Tools like Zapier and Make allow you
to run each individual step separately to test functionality and isolate any bugs or issues to a specific
step.
This step is pretty self-explanatory, but once you feel good about your automation plan - it's time for
implementation!
As you implement the automation, test each task to identify and resolve any unforeseen issues. Once
the automation is built, deploy it to a test setup, if possible, before rolling it out fully.
1. Summarization
Create an automation that summarizes content such as emails, meeting notes, news articles,
customer questions, etc.
2. Classification
Create an automation that automatically categorizes text or data based on predefined criteria, such
as sales lead qualification or customer questions
3. Extraction
Create an automation that extracts the information you need out of messy input data such as
customer onboarding forms.
4. Structured generation
Create an automation that generates content based on predefined templates or structures such as
marketing campaign briefs or client onboarding docs.
5. Freeform generation
Create an automation that can understand the input and draft an appropriate response, without
needing an explicit template such as auto-drafting email responses.
6. Translation
Create an automation that translates from one language to another, even from english to SQL or
Excel formulas.
7. Recommendations
Create an automation that can serve as a lightweight recommendation system to offer customers
personalized advice or handle complex routing of requests.
8. Sentiment analysis
Create an automation that determines the sentiment or emotion expressed in a social media post or
customer email, such as positive, negative, or neutral.
Go deeper on AI + Automations:
Get access to our full library of resources, tutorials and our
AI for Work Playbook at theaiexchange.com
The AI Exchange is the largest focused professional membership dedicated to empowering founders,
operators, and professionals on their journey to adopt AI. Founded by Rachel Woods, a former
Facebook Research Data Scientist and prominent online AI educator with 150K+ followers across
platforms. Learn more at theaiexchange.com