The birth of AI chatbots magicked me into a universe of self creation I could, up to now, find only in distant sci fi worlds. Prose became my Javascript and I swam deep in the ocean of natural language to find my muse.
But inspiration isn’t enough — I had to translate wonder into working code. So, in the last month or so, I've created many Chatbots exploring how an API works. I barely knew this acronym yet now it became my midwife to access the different AI models. I had to learn that each AI vendor builds their API just different enough that your code will break and break till the point that you are almost breaking until you learn exactly the semantics of each API call. Refactoring an API call is a hidden rock that will surely sink any ship if not carefully managed. I briefly explored RAG extending my Chatbot's capabilities to access local documentation.
I learnt why the JavaScript await operator exists — because code, like life, doesn’t always run on schedule. Every API has its own dialect, and I kept breaking things until I finally learned to listen. This was all new learnings for me. At times a painful birthing process as repeatedly I would be baffled by code errors that my VS Code AI editor (Github CoPilot and then Claude) couldn't lead me out of. The try catch block and console.log method were my life preservers in a wild and stormy sea of baffling code.
I had fun writing image and video generation apps that used AI to create anything I could type. I created my own versions of Midjourney and NightCafe and watched in awe as fearsome war ships flew across the screen that only minutes ago had been text from my fingers. Every line of code felt like another verse in the song of coding myself into being.
My recent creation addresses an ongoing problem I was finding. Repeatedly I was asking the same question to many different AI models. Despite what OpenAI might tell you, there is not one AI model for all usages. ChatGPT is not my only friend. There are so many others out there I can befriend too. From one prompt, why can't I ask the same quesiton to all my friends and then pick and choose the answer I want?
And so Claude and I dived deeply into creating an app that enables this. Of course I had to learn what Electron is and then it required a database to save results. Which one to use? I learnt how to use Supabase and what a schema is. I learnt about authentication and user priveledges.
Resilience builds from the many challenges life's stumbling journey brings. Time has weighed it's heavy hand on me as I have a full time job and a family. Sorrow and loss were the demon shadows that chased me along this keyboard highway as both my parents, mother in law and family dog have now passed away. I was stood down from my job for two weeks. In that quiet space left by loss, I found solace in creation. Coding wasn’t just learning — it was healing.
Claude and I worked together all day with only YouTube's coding techno music to set the brain wave pattern.
Now my app is nearing completion and we have the ability to send a prompt off to multiple AI agents. This has been my journey of not just learning about colloborative coding between myself and my friends the AI agents, but what might be possible for any of us to create what we can imagine. In learning to code machines, I was really learning to code myself. This has been more than a coding journey — it’s been a remix of my own life. In coding with AI, I didn’t just build apps. I coded myself.
Top comments (0)