2. System



MINIMUM VIABLE PRODUCT (MVP)
Playform started off as this—an MVP that was called Onerio, where any artist who was reached by word of mouth could freely try out two AI processes. These two processes, Freeform Dream and Define a Style, formed the technological aspects of the product, and everything else supplemental to these two processes were part of the user experience.

I was brought on board to the team shortly after the MVP was created to bring in fresh new ideas around where the product could go and to improve this user experience.


IDEATION
At the time, I was given free reign to re-examine the MVP, and many of my studies and tests around improving the user experience of this MVP were constrained with the design language and systems that were set previously.

Luke Cheng, who was directing me at the time, wanted us to go further and have explorations around this new way of creating. One metaphor that was used primarily the MVP-era of Onerio was “dreaming” to represent the training the AI needed to do to understand the input images and produce new images.


I came up with a few new proposals to target our needs at the time:
  • Visualizing different layers and ingredients for a project that influence its results, in a clear manner for artists new to AI
  • Creating a database of image sets to allow  for more easily accessible creation, first going in the direction of providing public domain historical art images
  • Trying out different metaphors for AI training, including “fermenting” and “primordial soup,” both which appealed to millenial artists but missed the mark for older artists on the platform
  • Testing, communicating, and understanding users’ needs around the education of AI

With our extremely small team, however, it was relatively unfeasible to create these custom experiences for any one specific process.

As our knowledgebase grew and as we conducted more experiments, there was a shift within the company itself that then wanted to create a platform that could encompass many more AI processes in a flexible manner, rather than just than any one process.

CREATING THE SYSTEM
The version of the system you see here is one of refinement and iteration. 

In our first iteration, we started with just one AI process on Playform and have progressively added more processes over time, to both balance our feedback loop with our users, along with the engineering load for an extremely small team.

We moved away from the metaphors of “dreaming” and “fermenting,” and rather toward language that is transferrable when talking to other artists who may be using AI — terms like “training” and “model.”

The most important jump from our MVP to this system, however, was the inclusion of a projects system. Previously, our users’ most prominent pain point was the inability to re-reference their projects, unless they had access to a confirmation email hyperlink. With the newer system, everything was finally bundled in one place.

In following iterations, we were able to add in:
  • Collections, which includes not only open source and public domain images, but also the ability for users to share their collections to the larger community. Again, this allows artists to spend less time scouring for images if they simply want to experiment.
  • Explore, which showcases exhibitions and opportunities provided by the Playform community, as well as location for artists to share single images and entire projects.
  • Processes, where we were able to educate users just enough to get the experimenting, avoiding the longform texts that were in the low-traffic, off-site docs.

As the sole designer for much of this new system, I often found myself needing to trust others. I would have to collaborate with our engineers to figure out suitable ways to solve our design issues and trust that they could optimize it in the way they knew best. I would have to trust the direction that the company itself was going in, like when the goals shifted during our Ideation process. And of course, I would have to trust our users, the artists on Playform, about the needs they voiced and adjust our product accordingly to make sure they were satisfied. 
Mark