The Rise of Devsumer: Or Story of the Lazy Coder

I wanted to take you on a quick journey through the past ten years of app development, mostly through the lens of my own personal experiences and highlight the trend of rising devsumer space and the lazy coder!

Phase 1: Build em All (2009–2014)

The story begins in Westwood, CA, in 2012. I was sitting in a coffee shop with a copy of Lean Startup and Lean Analytics by my side, building a mobile game called Berry Crush. This was one of the handful of apps I released that year, a combination of my love for all types of berries and the rise of mobile games.

Now the actual app is not as important as the tech I was using to build it. This was the year where Heroku was on the rise and the combination of deploying via git push and the shiny new design of Heroku made me fall in love. This was definitely an upgrade to running my own XAMPP and SQL servers, and I was enjoying my shiny new scroll bars for increasing the number of deployed servers. I still think Heroku’s design was one of the best of its time.

I built a beautiful backend in Rails that had an authentication layer and added some minor REST APIs for storing and retrieving data. ActiveRecords were all the rave at that point, and I was so happy that the read/write operations were all automatic and I didn’t have to write any more SQL insert queries! 🤣🤦‍ This combined with some request wrappers in Obj-c, unblocked the app and I was able to launch.

Of course, the deed wasn’t done without some analytics — and after thoroughly examining Lean Analytics, I decided on the events I wanted to track… and what did I do after? Of course, built more REST APIs! My custom made dashboard was janky, so I slapped on some JS libraries that showed beautiful graphs and voila, I got myself analytics!

Overall, this was a lot of work, basic blocks just had to be built.

Phase 2: New Shiny Tools (2014–2020)

Fast forward a few years, I was enamored by movement of financial data and decided to build an app for tracking flow of dividends. It was called Dividend Watch and it helped you track important events about your dividend stocks, mostly centered around how much and when you’ll receive your dividends!

This time around, I had a few more shiny tools! Parse took over some of the authentication and data storage layers and even better, I didn’t have to bring back my janky analytics dashboard! A new shiny tool called Mixpanel and it had these beautiful dashboards all for free!

In the meantime, I did a quick foray into AWS’s ElasticBean, Amazon’s answer to Parse and backend solution for app developers, but it was just nowhere near good usability. AWS’s documentations still reminded me of reading Microsoft’s thick documentation books in early 2000s. The elegance was just not there and I much preferred playing with Parse.

This was the infancy time for devsumer. As an engineer, I could pay $25/month to get an awesome analytics dashboards instead of building one myself, but the final product still had enough complexity to juggle for me to be a full time coder.

Phase 3: The Rise of Lazy Coder (2020+)

Fast forward, and everything from sending out emails, to sending out SMS, analytics, dashboarding to building full featured web sites has a tool that does it with little to no code. You don’t even need custom code to combine these together like legos, just use Zappier, and little zaps will pipe your new customer acquisition through simple pipelines, whatever that may be.

Still a bit early, but the rise of serverless code is mind blowing — no more servers to scale, no more micro-services all intertwined together orchestrating a maze of communication. On a serverless stack: here is a function, please run it, somewhere, somehow, and boom: firebase takes care of it! How amazing?!

Apple is requiring you to have a support web page? You could fire up VSCode and write a little React script, or why bother, use Bubble and build the full fledged support site in less than a few minutes.

I love this phase. It’s making me lazy — I want to be lazy. Let me focus on the harder problems. Do my customers value what I am building? Do I need to scale sales organization or does my CAC < LTV make sense? These are the problems that will make or break a business, not if I can write a cron job to send a reminder SMS to my customer in the morning.

My optimism falls short as the complexity of the needs go up, but this tech is so promising and it’s clearly indicative of where the future is headed. It’s elevating the abstraction layer at least 10x, making my life as an engineer equivalently easier!

So what does this all mean?

Firstly, more of the building blocks will get abstracted behind tools. I suspect things ranging from content moderation to a well crafted app onboarding flow will get abstracted away. Any major common problem encountered while building an app will be wrapped in a tool and sold as a service — after all, the cost of a junior mobile engineer starts at $100k, why not cash out $15 for an auto-generated onboarding flow?

Secondly, the rising tide of tech savvy knowledge workers hungry to build without coding will pull in more tools similar to WebFlow and Bubble, creating even more world-wide economic opportunity! Tools like Adalo are starting to pave the way for no-code mobile apps as well and soon most use cases in mobile will be easily built without any line of code.

Lastly, it’s a bit early to form a strong opinion on AI generated code, but the demo with GTP-3 based code gen solution is mind boggling and I suspect that either we’ll have comprehensive code completion solutions out of it or complete take over for cookie cutter solutions.