Product Design & AI
I look forward to revisiting this entry in six months to see how things have changed. Evolution is, and has always been, a natural part of life. Generation cycles are quicker, and now, so must we be.
Intro
At its most simple, product design is the understanding of a problem and the context the problem lives in, expressed through a UI.
A design’s “goodness” is then, a function of the level to which the problem and its context was understood.
Context composition is problem dependent, but things like, the competitive landscape a business operates in and the goals that drive them, user data and user goals, user attention type, customer acquisition methods, an existing design system, and a product’s codebase are all contextual elements shared by most problem spaces.
These inputs are some of the primary drivers of a product/eng team’s ultimate solution.
In the age of AI, the human moat is the ability to unify disparate information and process the composite. It is far from trivial to connect all points of context needed to achieve an output that is indistinguishable from a human, to an AI system. And even if you are able to offer exhaustive inputs, past a certain point, as the number of inputs increase, the reliability of the system is liable to degrade (often expressed through "hallucinations").
Until such a time when outputs are indeed indistinguishable, it is sensible to progressively load AI into the systems that make work possible.
Below are two applications of AI I have found useful for making design decisions.
Custom GPTs
Collaboration is an essential part of product design and solution delivery.
Though not always, the value of collaboration is typically commensurate to the level of understanding participants bring to a discussion. A richer understanding of the context the object of discussion lives in and the things it is impacted by, will often lead to better outcomes.
I contest that collaboration, at the time of writing, is still best reserved for humans, but powerful insights can be uncovered when thoughtfully collaborating with an AI companion that is equipped with context and a bounded intent.
I am able to realize such value through the creation and integration of custom, app-specific GPTs.
In Practice
The primary value of a custom GPT is contextual and instructional persistence. Once a GPT is configured, its purpose is cast, removing the duplicative prompting of restating context in each new conversation.
It remembers.
To ensure focused, productive conversations, my custom GPTs are segregated by the platform I am working on. Though segregated and distinct, I equip all GPTs with identical types of information. I offer the platform’s purpose, the broader context the platform lives in (including, but not limited to, the way it supports business goals), the types of users accessing the platform and their goals, key design system patterns, the idiosyncrasies that affect design decisions and trade-offs, as well as the explicit instruction for the GPT to act as a thought partner.
Without the overhead of ensuring context is remembered, or having to re-establish it, the focus of each conversation can regard a specific user, system, flow, or business problem that is intended to be solved through design.
User Intelligence Layer
In addition to a companion thought partner, I have found considerable benefit in leveraging AI and automation software to automate the collection of user information before synthesizing it.
Through Zapier, I am able to ingest many different types of information - call logs, surveys, user session tracking data - extract certain information, store that information, and then analyze all information for trends and patterns.
The beauty of this system is its perpetual currency - data continually flows, information is continually extracted and stored, and trends are continually updated. It is not a perfect replacement for a dedicated research role, but it is a powerful system that goes a long way to ensure users and customers maintain a seat at the table of a product team.
Conclusion
Points of leverage in AI, with specific respect to design and product delivery, will undoubtedly change in undoubtedly short order. In terms of practical implementation of these new tools, these are two examples are relatively low effort and low disruption, that yield value very quickly.
I am excited to see what else can be integrated into the systems that drive my work.