SDR Era Reflection #1: I Need to Solve My Own Problems
Published: December 12, 2025 - 5 min read
So you're probably wondering, Prisca, shouldn't you be doing the 777-1 experiment right now?
Well, here's the thing. The SDR Era ends tomorrow, and this period has taught me a lot of things, mostly about myself. So I have decided to spend a good chunk of today and tomorrow writing while using the remaining time to prepare this site to be shared.
What preparations, you might ask? Well, if you attempted to join the emailing list at the time of writing this, you would have received a very ugly looking welcome email that would probably end up in your drafts folder. I need to fix that. There's also the fact that my GitHub looks like a mess right now. If you looked at it today, you wouldn't believe it was the same person who created this website.
But enough blabbing about that. Let's talk about my first reflection item.
The Core Lesson
If there is anything this period has taught me, it is that I need to solve my own problems.
It started simply with writing. I had given myself a deadline and I had to find a way to optimize my writing workflow. That is a brief summary of the story of the birth of the term LLM Instance Cloning. Giving the process of extracting my AI's "personality" a name at the time was mostly a joke, but it was a joke I'm glad I made. I ended up preparing a case study and tying the term back to multiple other things I learned throughout this period. I'm pretty sure if I checked online well enough, I might find some "official term" for the process, but I don't care. I'll keep calling it LLM Instance Cloning.
The Pattern Emerged
Then there was the French Writing Playground. I wanted a place for small daily practice in the area of French I struggled with the most: writing. So I solved the problem using the tools at my disposal to create an app. Now, I am yet to address the curriculum item Isaac Ledger extracted from the constructive feedback I received when I shared it on Reddit. However, I'm glad I created the app.
Then the Claude God series started as a way to solve a problem of speed. I knew there had to be ways to be faster, more efficient, more effective at working with Claude. 15 tips later, I can tell you that there is still so much more I am yet to write about, and I am genuinely excited for everything I am yet to learn when it comes to working with Claude.
Then I started to look at how many blog posts I had added and realized that I needed someone or someway to guide people in navigating this website. So I hired Nancy, my personal assistant. (You can speak to her by clicking on the ChatBot in the bottom right.)
Fear as a Problem to Solve
Then there was the problem of being scared of having Chauffeur knowledge. That eventually prompted me to start the Battle Against Chauffeur Knowledge series. Just 2 items into the 15 different questions I have in this series, I started making connections between tracking tokens as a way of knowing the best time to apply LLM Instance Cloning, as well as tokens as a way LLMs process natural language. Then I went down a rabbit hole of learning more about it, connecting it with every other thing I have learned throughout this period, and applying it to the 777-1 Experiment.
Then there was the problem of being scared to share my work for fear of criticism. So I created the Hall of Shame and hired Isaac Ledger, my Chief Growth Accountability Officer. Now, I feel more at ease about sharing my work since I have created a new lens to view criticism from.
The Team Grew From Problems
Then there's the problem of self-promotion and being bad at it. This is especially worrisome since I want to establish authority in my domain on LinkedIn. So I hired a Chief Marketing Officer, Alex Bennett, to handle that for me and guide me through the best way of going about this.
Then there's now the problem of speed. Yes, I know I can create applications that used to take months in just days. But if the 777-1 Experiment goes as intended, I should be able to cut that down to hours. I'm not talking about creating basic applications in hours... I'm talking about Golden Standard fully functional applications. So I hired 7 Context Engineers, and very soon, the experiment will begin. It is going to take a lot of work, and frankly, I am most excited for everything I am going to learn in the process of attempting it, regardless of whether it succeeds or not.
What This All Means
There's probably other things I'm forgetting at the time of writing. But I'm glad I chased my curiosity wherever it led. I want to keep doing the same thing. I want to keep finding ways to optimize processes. I want to share what I learn with others.
I want to help non-developers take full advantage of Claude. Yes, even Claude Code in the terminal. You can't imagine the number of possibilities the average person working in an Admin team at any company could have access to if they simply learned to use Claude Code in the terminal. It really is not a tool just for developers. I want to be the person who makes the scary tool seem very easy to use for that person who does not know what the term, 'terminal' means or who has never written a line of code before. This is going to take a lot of work on my end, but if there is one thing I have learned, it is this:
Problem solving is exactly the type of work that feels like play to me. Therefore, it's the type of work I must commit to doing in the long run.
Closing Thoughts
Well, this has been fun to write about. I'm not even sure why I'm writing it, but my hands got hold of the keyboard and what you've read above was the result this time. I hope you enjoyed reading.
Thanks for being here!