An attribute of a highly successful Product Manager is the ability to effectively communicate their vision. What the product needs to do, how users and systems will interact with it and when it will get to market. The usual tool for the majority of PMs is some combination of PowerPoint, Word, Excel and email – the usual business tools.
The most impactful vision discussions have a tractable component that enables the audience, whether customers or partners or internal teams, to really see what the PM is imagining. Knowing this, most PMs will work with a UX tool like Figma, Balsamiq or Sketch.com to create an exemplar walkthrough set of screenshots.
At the end of that process, we have a visual representation of the workflow from business problem to solution. The concept for innovation, not actual innovation.
GenAI as Innovation Enabler
Generative AI technology has the potential to change how Product Managers, and anyone else, is able to work with advanced applications and technology to move from vague UX mockup concepts to actual innovation. Not to the level of enterprise grade productization, but definitely to the “you can press buttons” demo that has actual data processing capabilities.
Meetings, the urgency of the inbox, hot customer issues, etc. etc. all pull PMs away from being able to build out in more detail their innovation concepts. Which can unfortunately lead to less than compelling visuals and bumpy value-proposition communication. Solving that problem – allowing for faster innovation of product concepts to create high value demos or concepts means every highly successful PM will have to add “prompt engineering” to their list of skill.
Conceptually, this ties back to my earlier writing about the future of LLMs being micro (link). The current iteration of public LLMs is trained on generic data, so the output is generic too. It’s for this reason that the first attempts to use GenAI to write Python or code snippets in other languages was received with derision from experienced developers.
Microsoft has recognized this gap and addressed it with their latest version of CoPilot and it’s integration with GitHub for developers. What Microsoft is doing is laying the foundation upon which the business leader facing capabilities will be built. The future will be the GenAI interface for Product Managers to simply ask for what they want to have, and the system will produce a result that, at least surface level, does something.
Getting There from Here
Getting to the point where we can speak to the system and it reacts as we want is not a straight line. There will be hiccups and false starts, but the outcome is clear.
Leading developer tools and business applications will be implementing highly tuned GenAI capabilities that understand how to translate a loosely described business ask into a prototype of an answer. And that output can then be further refined until the Product Manager is satisfied with the result.
We can see some first iterations of this strategy in what Wix is doing. But even that is very much in the infancy of what’s possible.
Wix or WordPress or others will (if not already and I just don’t know about it) be extending this concept much further. And we will see it with developer tools as well, which is where together how PMs innovate is on the cusp of being forever changed.
For example:
- Users will be able to say “build me a blank website at foo.ai that is similar to xyz.com” and the system will know exactly what next steps to take
- Opening a code console and describing what you want done. “Open a dialog box to browse a local system. Then let the user pick a drive. Then look in all the files in that drive for any Word documents and list their names on my foo.ai website in a new page called “show_files.”
At the end of the process, the PM has a system that is mostly functional and highly demoable to show “art of the possible” to their stakeholders. And the underlying technology is GenAI with a highly tuned LLM.
Published 26-Sept-2023