GUACAMAYA

Origin

While writing the novel for the soul-kernel project with the help of AI, I came to understand the possibilities and limits of that form of work. From that exploration emerged tools and ideas that I later used in another book.

With the upcoming AWS Summit in Mexico City, I decided to share an idea through a publication and developed GUACAMAYA to support the process. The result was When the Toaster Replies, signed as Paynalton.

Work Process

The thesis of the book and my editorial practice share a principle: using AI as an additional cognitive component that expands human capabilities. I retain authorship control over the work and the decisions about what I want to communicate.

In my workflow, I distribute tasks between local and external agents. I use local resources for tasks like images and translations, which can consume a significant amount of processing or tokens, and external agents for tasks like writing and idea formulation. This is a work distribution approach, not a universal classification of the difficulty of those activities.

The experience also contributed to the origin of LORO, an additional layer of memory and context for agents. PERICO applies that line of work to software development. GUACAMAYA focuses on the editorial application, and the tools that helped produce the published book are now returning to the literary project that initiated the initial exploration.

Status and Next Steps

Currently, I use GUACAMAYA to support the creation phase of a personal literary project with external agents. When I conclude that phase, I will implement and refine the editorial processes that still remain.

The tool needs more templates, testing with other types of books and security testing. My personal opinion is to complete at least ten books before considering it a finished software. That number is a future validation goal, not a result achieved.

Results and learning

The first result is a published book. The experience allowed me to study how to maintain context, distribute tasks and keep an authorial direction during a complex process. The evolution of GUACAMAYA is based on using it on real works and learning from its limitations.