Recently, I did something that I should have started doing a long time ago, yet I’m still pleased with myself for doing it now. I saw a need for something in my own life and made it, albeit using the desire as my first experiment pair programming with AI. I’ve been on and off focused on working out, yet remembering to write down everything I eat in the same Notion spreadsheet would get tedious and I’d eventually fall off. Seeing the ease of using Alexas to turn on and off the lights, I wondered if I could do the same for food logging. A lot of devices I had included connections to Fitbit, so it quickly became my main health tracking app. Then I wondered if I could get Alexa to log to Fitbit and an idea for a side project was born.
https://github.com/nickhattwick/nickate/tree/main
How is it that after five years into my coding career, this is the first real side project I feel I’ve taken on. I’d be lying if I said AI wasn’t a part of it, I’ve been wondering for awhile what it’s code generating abilities were like and wanted to see what it would be like to work on a project where I use it fully, starting from scratch. Even before the idea struck me I was thinking of starting a “side project” to test out AI. And then the desire to log food hit me again, and then I got sick for a weekend, and then Nick Ate was born, using ChatGPT as a coding partner.
ChatGPT was generally quite right and a bit wrong on most of the things it suggested, telling me to set up a Fitbit app and Alexa skill, though often not fully accurate on the site layouts. Still it was sufficient for getting started with creating both, as well as the Lambda function. For testing, I created the Fitbit app first, then the Lambda, tested the Lambda, and finally the Alexa integration, initially having it log 1 of the default unit for the first item returned from the API and then providing controls for serving size as well as selecting another item.
Copying the code back and forth was interesting as well, since generally it would provide close to working scripts with some issues such as variables being referenced before being assigned. Sometimes it would be able to debug, other times I’d need to make changes before continuing. Not having worked with Alexa skills previously, it’s knowledge of accessing the objects passed from Alexa and using the Alexa Skill Kit was surprising, even now it would probably have been quite a learning curve to start that part from scratch using documentation alone.
So, AI is a huge time saver but the process is far from instantaneous. I’ve probably spent close to a week in total time (spare weekends across a couple months) getting this to its current state which has actually been helping me out quite a bit and am eager to keep testing AI and building out more tools to assist me, initially in health tracking and more later.
For now, I’ll probably leave Nick Ate in it’s current state for awhile with setup steps to deploy for anyone curious and begin working the next app to start using OpenAI’s API to summarize my health data in the form of a skill I’m initially calling Nick Life. Honestly, working on this was quite fun and I found myself having mixed feelings on working with AI. There’s still a lot of manual coding I find myself doing, yet it provides a speed up where I can see results in side projects a bit faster. While this initial code was made with ChatGPT assistance, I’m planning on diving into Github’s AI powered developer tools for my next addition to my fitness/productivity tools and continuous learning with AI as I was definitely a bit nervous seeing the AI automatically handling issues in their demo from Github universe and am very eager to try out their VSCode extension.
For now, check out Nick Ate on Github.
https://github.com/nickhattwick/nickate
While I’m adding more features for now, I may polish this up a bit further and make it easier to deploy in the future. In the meantime, I’ll keep working on features and post my findings.
Leave a Reply