A Challenge: Automate one PM task in just one week
A step-by-step guide to automating PM tasks like a pro
New to automating PM tasks with AI? Steer clear of the all-or-nothing mindset.
Too often, I see PMs trying to automate too much.
They aim to fully automate every aspect of the product lifecycle. From writing requirements to building roadmaps to creating GTM strategy from the get-go.
Unfortunately, the “all or nothing” is the worst way of approaching AI.
It’s like running a marathon without ever having jogged around the block. The goal is too big, and we get discouraged too quickly.
Try automating one small time-consuming task at a time for a better approach.
Here is what I would do:
Step 1: Identify one micro-task
Start by reviewing your daily activities to spot repetitive, burdensome, or time-consuming tasks. Anything that makes you cringe.
These are prime candidates for automation.
For example, one of the PMs I coached detested writing release notes. He had to produce multiple versions of the same note for 3 different stakeholders every two weeks. For him, it felt like a never-ending cycle of repetitive work.
Next break it down into smaller, more manageable micro-tasks.
Regarding release notes, we split the task into three distinct micro-tasks, one for each stakeholder group. This makes the automation less daunting and more approachable.
Step 2: Research best practices
Once a task is identified, it is tempting to start building your prompt.
Don’t!!
AI offers an opportunity not just to automate but also to improve the quality of the output. Take time to discover what good looks like. Research industry best practices, effective formats, and established methodologies.
For example, in the case of the release notes, a quick search online provides us with lots of best practices and top-tier formats (e.g. Intercom, Invision, Hubspot, etc.).
Step 3: Determine the right inputs
After defining what a good output looks like, the next step is to identify the required inputs.
For this, you can leverage the AI itself. Put the desired outcome in the chat box and ask, “What details do you need from me to create this release note?”
If the AI suggests 10 to 12 input items. Consider, narrowing it down to just a handful. It will save you time further down the road.
Step 4: Craft your prompt
Now we finally get to the “meat”.
For this, I use my 5 key framework. It is a prompt structure that is designed to generate high-quality responses right from the start.
The structure includes:
Assigning the AI a role
Providing the right inputs (from step #3)
Giving specific instructions (including the best practices from step #2)
Specifying the desired output format (again from step #2), and
Providing examples
Here is what we created for generating customer-facing release notes:
You can see the entire chat here.
For more examples of a well-structured prompt - check this out. And if you or your team want to learn how to write one so that you can automate any PM tasks, consider joining my bootcamp.
Step 5: Measure value
Many overlook this crucial step, but I consider it a cornerstone of the process. And I encourage my students to do the same.
This step is all about measuring productivity.
Here are the 5 things that I look for:
How much time did the AI save me (hands-on keyboard time)?
What was the improvement in velocity (i.e. elapsed time)?
Was the output in the right format?
What was the quality of the responses? Did it do 60% - 80% of the work? (hint: shooting for 100% is’’t realistic)
The ultimate test - what did my stakeholders think? Did I make them happy?
Step 6: Iterate
Your prompt is not going to be right the first time.
I have been helping PMs automate their tasks for the last 1.5 years, and even I never get it right on the first try.
So look at the above metrics and adjust your prompt as needed.
Maybe your format is off. Or you need more precise instructions. Or that you are not providing the right inputs.
Your goal is to get the prompt to a point where you can make it repeatable. This way, the next time you want to complete a PM task, you have to change the input and run the prompt through AI.
Keep adding one task a week.
Once you automate one. Then move on to the next task.
If you do this once a week. At the end of the year, you are going to amass a library of 50+ productivity-boosting prompts.
Now you are cooking :)
Start simple, stick to the above steps, and grow from there.
Happy building!!
Whenever you are ready, here are 2 ways that I can help you:
#1. AI for Product Managers Bootcamp (public): An intense 3-week bootcamp, where PMs learn how to leverage AI to save time, deliver top-notch work, and accelerate velocity. The next cohort starts 4/3.
#2. AI for Product Teams Bootcamp (private): Private version of the above bootcamp. 100% customized to your organization. Includes hackathons, coaching, mentoring, and executive playbacks.