It’s quite the impressive result when you finally solve a complex prompt engineering workflow, but it’s also generated a ton more work for me…
Just when I thought I had finally neared completion of editing my book, I also finally achieved full success on my GenAI manuscript editing workflow, and damn did it work. It’s both an amusing moment to solve a series of problems that unlock a huge capability, but also sort of frustrating to realize how much writing/rewriting I still have left in my book. The results of both of those situations are ultimately very positive for the book and all my future GenAI workflows. For context, my previous editorial workflows were generating maybe 1-2 pages of notes, that were mainly just resulting in positive reflections of the current content with a few suggestions for content improvement. The new workflow breakthrough yielded over 8 pages of extremely detailed and high-quality feedback that was fantastic! It really was one of those magical “AH-HA” moments of both achievement and realization.
The realization was that I knew the concept that I had been working under for aligning my workflows with instructional frameworks was correct, but putting the missing pieces together to actually achieve amazing results was something else entirely. The key breakthrough was an idea I started working on about a week ago, involving collecting analysis frameworks. The core concept was if you had many different ways of thinking about (solving/describing) a particular task or problem, you could change the underlying task from one of problem solving to instead optimum tool(s) selection. LLMs are much better at determining the best answer from a given range, than just generating the correct answer from scratch. So, I researched/collected/documented 13 analytical frameworks, and had the model determine the best framework(s) to use and in what order for the overall task at hand (in this case literary editing). I used the overall result to subdivide the workflow process into fiction and non-fiction workflows, and went about the execution for my manuscript as a test….and SHAZAAM!
The really impressive thing about this approach is the way I solved this task is the way I can solve many tasks, and I can’t wait to use this approach for more use cases…stay tuned.
