Thirty days of agents: motivation

Anyone who loves the beach must at some point in their life visit Rio de Janeiro. I have no doubt that Ipanema beach has something you want. Be it sun, surfable waves or refrigerated beverages. And let's not forget about the phat view. As the sun sets behind the Vidigal favela, you can take a moment to reflect on the day that has now passed. Your mind slips into a dream-like state and you feel grateful and warm until suddenly you are ripped out of lucidity by the fresh yet sterile glare of the beach floodlights.

This transition of light - from the sun's natural rays to artificial, mechanical light - turned out to be a fitting metaphor for the theme of this blog series. After all, we software engineers are witnessing in real time how artisinal hand-written code is being replaced by probabilistic streams of tokens.

When I found myself jogging under these bright lights in Rio a few weeks back I was listening to a podcast about Openclaw, the agentic development framework that has taken the tech world by storm. It was quite motivating to hear how the project's author Peter Steinberger had rediscovered his passion for coding whilst developing Openclaw as I too had been struggling to see the beauty in work I was doing. The last time I found myself in this situation was back in 2023 not long after the release of ChatGPT (wow - those years flew by!). Back then I was working as a data engineer in my first proper coding job since teaching myself to code during the covid pandemic. The antidote to my growing disaffection for my first language python was the discovery of my second language - rust. Having attended Pycon that year in Berlin I noticed that rust was becoming the language of choice for anyone who wanted to rewrite python libraries to make them faster and/or functionality better. One prime example was the evolution of the polars (rs for rust, you dummy) library which was starting to challenge the data wrangling hegemony that pandas had over python devs and data scientists. Back then it felt as if the future was rust and it was rewarding to find a way to ride that wave.

Agents and frameworks to manage them represent a fundamental rewrite of that future I had imagined. One the one hand, they represent a "democratization" (to borrow the vernacular of tech leaders in the 2010s) of access to software. Even bros in Berghain will tell you how they developed an app with Claude. For experienced developers the offer huge productivity gains and bootstrapping companies has never been easier. But importantly they also pose a risk to the industry of software at large:

1. Large tech companies will be able to maintain monopolies with fewer engineers
2. Engineers who aren't willing to embrace significant change will get left behind
3. Private equity firms with big bets on pre-agentic enterprise software fear loss of market share (white collar jobs replaced by agents)

And so in many ways it is a bittersweet moment for me, accepting that I must and therefore ought to want to change. I believe passion, interest and knowledge are not virtues born out of nothing. You have to work hard to cultivate them and that's why I want to share this journey with the wilder world. Our first stop is going to be getting Openclaw up and running. Watch this space.