This week's tip: How to use "AI-Templating" to break down workflows into goals and tasks.
Today, I’m going to walk you through how to use AI-Templating before you even start building AI-driven workflows to ease your job.
The reason that Templating is so critical, is that it helps validate that the workflow actually is a good fit for Generative AI.
Here’s what happens when you don’t use Templating:
"...jumped into prompting immediately and at the end, spent more time for back and forth editing and prompt refinement than it would have taken to actually do the task without AI."
Unfortunately, this is a common occurrence.
So many professionals I talk to spend countless hours and energy on prompting for tasks without ever validating the actual goal of the entire workflow.
Let’s make sure that doesn’t happen to you.
Instead, I want to walk you through the 6-step process I use to streamline my workflows before even using ChatGPT at all.
Here’s how to do it:
Step 1: Choose a workflow.
There are tree categories of workflows that are ideal for templating:
- Workflows that consume disproportionate amounts of time
- Workflows that are repetitive and do not require creative input
- Workflows that are simple yet critical for day-to-day operations
Step 2: Get clear about the goal.
What is actually the goal behind the workflow you want to reach?
In other words: If the workflow had a title what would it be?
Examples:
- Consulting: Preparing for a client meeting.
- Sales: Drafting personalized outreach messages.
- Operations: Creating bullet points form a meeting.
Step 3: Break down the goal into its tasks.
What single tasks are needed to solve this goal?
What should be the outcome of the single task?
Example:
- Meeting recording, to transcript, to action items, to email
Step 4: Create prompts for each single task.
Now that you have identified the single tasks,
go step by step and create a prompt for each task.
Step 5: Put it all tasks together in a prompt chain.
A prompt chain refers to taking:
- the output from prompt 1 as input for prompt 2
- then output from prompt 2 as input for prompt 3 and so on...
Step 6: Run the prompts and iterate then revise regularly.
The final stepp refers to:
- testing the whole workflow,
- iterating and improving the prompts
- checking the outcome to build trust in AI
Congratulations, you have added an AI-driven workflow that makes your job easier.
To stay informed about AI for Business:
Follow & Subscribe: LinkedIn & Newsletter
For feedback, questions, ideas Contact Me
Comments