Since the first days of building Sublayer, we’ve been asked why we decided to build a new product management app instead of just integrating with the existing ones. It wasn’t a decision we decided to take on lightly – the first proof of concept as we were exploring the idea was actually integrated with Pivotal Tracker (demo video here), but the further we took the idea the more it became clear to us that even though we’re starting with the product management interface metaphor, where we were going to end up was going to look much different. We needed the freedom to go down directions that just weren’t available through other services’ APIs.
But Why Start With Product Management?
The simplest answer is that the UX of the product management app is already a pretty optimized UX for building and changing software over time. You can even think of it as a prompt builder – the person writing the stories or specs fills in the state of the world, some motivation and description for the change, and the expected output sometimes in the form of acceptance criteria. If you’re a product manager, you’re already a prompt engineer, you’re just prompting humans to change your software.
We believe the way that we’re going to collaborate with AI to build software is going to feel a lot like product management does today, with the PM role shifting to be a little closer to engineering and the engineer role shifting to become something a lot closer to PM than it does today. But, if that’s all it was, the existing solutions out there would easily be up to the task of integrating generative AI and there would be no need for anything new.
Engineering as the Bottleneck
As Goldratt taught us in The Goal – any improvements made anywhere besides the bottleneck are an illusion. PMs writing stories faster, designers designing faster, customer interviews summarized faster are all great, but none of it will lead to increased output and to increased revenue. The current system is so ingrained in the way we think about building product that until someone breaks the bottleneck of engineering, it’s hard to guess where it is going to move to next. This is the only step in the process that matters right now.
In most, if not all, product organizations, engineering is the bottleneck. It is the most expensive, and slowest step in the process of delivering software. Everything in these organizations is set up around the delivery cadence of and capacity of this bottleneck. With preliminary numbers coming out around usage of Copilot at over 50% increased task completion speeds [1] we’re only seeing the tip of the iceberg for how this is going to transform the work of software engineering. With the ideas we’ve shared around Promptable Architecture we see the possibility of breaking the bottleneck of engineering completely, and are excited to see where it moves to next and get to work tackling that one.
An Opinionated Process
The other thing we touched on in Promptable Architecture is that the way we organize is going to need to change. The current products on the market are flexible and allow their customers to bring whichever process they like to developing products, and while we all laugh at the iterated waterfall / agilefall monstrosities that some companies have created, the variation in process at the current level just won’t exist in the future. When the risk of getting it wrong means a task taking orders of magnitude more time and money, the winning workflow tools will be the ones that make it hard, if not impossible, to do things the wrong way.
This is why we need a new product management application. To guide and teach users on the best ways to get the massive benefits available in using generative AI.
At Sublayer we’re obsessed with the product development process, and are extremely excited by what we’ve already seen with our product. If this post has reached you, sign up and try our product out, or join our discord and say hi!
