The promise of AI, at least for now, has been greater efficiency, a freedom from boring redundant tasks and the ability to free up time to do “more important” work. At a glance from those not doing it on a daily basis, there seems to be a lot of this type of “waste” in product management and product development in general. Long meetings, documents nobody read, frustrating alignment efforts that seem to lead to nowhere are just a few examples of this perceived inefficiency. It’s true. Some of these activities are wasteful and AI is well-suited to optimize and automate them. But others are not. The problem is that, at a high-level view it’s hard to distinguish one from the other.

In a recent engagement, we worked with a client on an org-wide OKR rollout. These are, at their core, goal-setting conversations. Goals are sticky topics because they ultimately end up impacting compensation, promotion, prioritization and overall business impact. The discussion to arrive at a shared goal is always going to be messy and inefficient. And this is exactly how it went with our client. One manager who was participating in these conversations kept trying to cut short the messy alignment conversations. They didn’t see the value in continuing to discuss these ideas and pushed to “get to a decision” as quickly as possible. Here’s the thing though, it was exactly in those conversations, the push, the pull, the compromise and the idea creation that the clarity and alignment the team was seeking was happening. It was the argument itself that got the team to agree. 

From the outside, those meetings may have seemed inefficient or wasteful. This manager certainly thought so. Had they been successful in cutting these meetings short the teams would still have arrived at a set of OKRs. But they would have been goals no one believed in or was committed to. We had a handful of groups working on their local OKRs and then coming together to bring the entire business unit together. The friction this manager was trying to remove was where the work was actually being done.

How AI risks removing innovation and alignment from your teams

AI is working on at least two fronts when it comes to visible waste reduction. On the one hand, organizations are looking to reduce unnecessary meetings, discussions and conversations with greater agent autonomy and access. The other efficiency is document creation. Anyone in your company can now create a plausible roadmap, PRD, OKR set, etc in an afternoon. There was a time where the creation of the artifact meant a decision had been made. Now it simply means a prompt was executed. 

Both of these efficiency efforts remove that messy friction from the product development process. By removing the discussions around the creation of the artifacts, the decision-making process is left largely in the hands of AI and the person doing the prompting. However, it is exactly in this mess that ideas get debated, manipulated and augmented to become better than their original version. Iteration and collaboration drive the creativity and innovation teams need to grow their business impact. Seeing that process as wasteful and using AI to remove it presents a huge risk to your company’s competitive advantage in the market. 

What can product leaders do to keep AI from automating the “good” friction away?

Let’s be clear, looking for process efficiencies is never a bad thing. And if we have new tools to help us with that, fantastic. Let’s use them to that end but you should use your new AI tools like a surgeon rather than a butcher. We want to make surgical cuts to the things that can truly be automated rather than amputating whole chunks of work that have value within them. Your first task is to differentiate between these two types of “mess” or waste.

Once you have a sense of where conversation, discussion and argument bring value in your process, task your teams to do the following:

1. Become better facilitators. Facilitation is even more core to good, modern product leadership work. It can’t be handed to whoever happens to be running the workshop, because the facilitator is the person deciding which friction to protect and which to cut, in real time. This judgment call ultimately affects product quality. 
2. Push your team to make decisions explicit. The simple act of creating an artifact or document no longer comes with the assumption that the author of that document made a series of explicit decisions. AI made that difficult. In these “messy” meetings, product managers need to push for the why behind each decision and to document why each one was made. That’s the purpose of these conversations. The assumptions behind the decisions will eventually be tested in production. 

How to determine which meeting is waste and which is messy but useful

Deploy your AI tools to find efficiencies in your workflow. However, don’t let the AI make the final decision on whether to cut a part of the process or not. Instead, with each proposed efficiency, ask yourself, “What does this step in the process decide for us?” If the answer, is “not much” or “it just moves information from X to Y”, this is a good candidate for automation. However, if the answer is, “this is where we decide whether we do X or Y” then that step remains. AI can help you prep for it but it shouldn’t replace it. It may also be beneficial to work with your team to identify the key decisions they need to make as a team, where those decisions happen and ensure those parts of your process don’t get automated away. 

Given all the various types of artifacts, documents and conversations your team has on a weekly basis, where does human interaction make the product better, more innovative or impactful?