Do you use AI tools in your role as a Product Manager?

In a world where generative AI tools have a bad reputation for hallucinations, it can be easy to dismiss the tools as not being very useful in a product role where accuracy and clear communication are essential.

As product people, the value we bring is acting as a communicator. We are the people who sift through the fire-hose of information, and pick out the important stuff to translate into system requirements.

But what if the AI tools could remove some of the drudgery of document creation, and get us 80% or even just 50% of the way there towards a business case or an epic definition in a matter of seconds? That would free up time for us to engage with customers, think more strategically, and give feedback to the development teams.

I’ve been sceptical about the quality of output that generative AI can give, and have hardly engaged with any tools that are available. That’s until recently, when someone stepped through how they could be applied to product management, and I must say, I’m pretty-much sold that AI actually could be the future – not just of product management – but of any role that involves critical thinking.

AI is my intern

What do they say about data? Put rubbish in, and you’ll get rubbish out. The same applies to generative AI tools. In the past I’ve tried out the tools by posting in a single-sentence question – and I’ve not been impressed by the quality of what comes back.

But imagine for a moment that AI isn’t an expert in a field who can provide perfect answers – but instead is an intern who’s still learning on the job. If I ask an intern to write a business case for me, they will make a reasonable job, but will most-likely require additional coaching and feedback to refine the business case to be better.

That’s how we should view generative AI at the moment. We can’t expect it know exactly what we’re looking for just from a simple ask. Just as with an intern, we need to give context, examples, and structure to our requests. We need to be clear about what output we are expecting, and how it should be formatted. And then the tools suddenly become more useful.

It won’t get us 100% of the way there. We still need to take the outputs, validate it and add to it – but it can certainly speed up the process, and help get past the inertia when staring at a blank page!

From workshops to roadmap

Let’s say that I’ve just finished a 2-hour workshop with my customers to talk about their adoption of a new feature. The AI tools in Zoom have already provided a full transcript of the conversation, and maybe even a list of actions identified on the call.

But what if I took the transcript of the call and, with a bit of clever prompting to the generate AI tool, asked it create epics for the feature enhancements identified in the workshop? I tell it the format I want the epics in, and maybe (if I have a suitable connector) even get it to populate Jira with the details.

That’s just saved me a lot of grunt work, and I can concentrate in validating and refining those epics with the dev team.

Then if I need to prioritise a set of epics, and I feed in the epic details and strategic goals, and ask the AI to stack-rank them based on alignment to goals and value for customers, and it puts them in a roadmap for me.

This might all seem like a vision of the future, but this is available now – and the chances are that some of our colleagues are already doing this.

But the magic doesn’t come from the AI itself – it comes from the quality of the prompts I can write. Just like with my instructions to the intern, I need to think about how to guide and coach the AI to produce the results I need.

Sharing is caring

Learning how to prompt the generative AI is a new skill that everyone needs to learn. And like with any skills, we as a product management community can help each other out.

As we experiment and refine our prompts, we can share them with other PMs in our organisation or wider community. We can learn from each other about techniques to get better results. And we can start to build up a library of prompts that can be reused across the team.

And with a pre-defined list of good prompts, we can get the generative AI to do a lot of the heavy lifting, and super-charge the process of documenting requirements.

It’s never going to do away with humans. We still need people to impart their domain knowledge, and to validate and correct wrong assumptions. But with our AI interns working for us – as artificial employees – we can boost our productivity and help our teams move with more agility and speed.