2023 Year in Review: LLM Integration, Payments, and Go

I’m writing this on Christmas morning with coffee that’s too hot and a year that went too fast. 2023 was the most professionally intense year since I went through a founder program in 2019 and started figuring out what kind of career I actually wanted. This year I found out.

Several Teams, One Thread

2023 took me through several teams. Some of the work was LLM integration in Python. Some was Go backend work on payments, the kind of work where correctness is the entire point. Every line of code I touched had to be right because the alternative was someone’s money being wrong. The rest was back in Go, working on LLM deployments.

The thread through all of it was the intersection of production systems and AI. The question that kept coming up around financial data: can AI help people query transactions in natural language? Catch anomalies? The answer, frustratingly, was “sort of, but not the way you think.”

AI around money is an engineering challenge that touches trust, auditability, and regulatory compliance at every level. You can’t bolt it on. I spent months thinking about how to make AI features that are safe enough for financial data. I’m still thinking about it.

The AI Explosion

I don’t need to tell you what happened in AI this year. You were there. But living through it as someone who builds production systems was a specific kind of experience.

January started with everyone experimenting. By March, teams were asking when they could ship AI features. By summer, the questions changed from “should we use AI” to “how do we make it reliable enough for production.” By November, OpenAI DevDay reset the baseline for what the platform provides out of the box.

The speed was genuinely disorienting. I wrote a blog post about agent architecture in September and parts of it felt outdated by November. I spent much of the year building RAG pipelines, and in November the Assistants API made half of that plumbing unnecessary. The ground shifted faster than I could blog about it.

What I learned: the teams that did well in 2023 weren’t the ones who moved fastest. They were the ones who picked a lane, built evaluation infrastructure, and iterated with discipline. The teams that chased every new capability announcement ended up with half-built features and no quality baseline.

Reflections

This year cemented something I’ve been discovering over the last few years: I like going deep on a problem, building something that works, and then moving on to the next challenge. The variety keeps me sharp. Working on payments and LLM integrations, thinking about security the way cyber-defense exercises taught me to, contributing to open-source Go projects: the breadth makes me a better engineer on each individual project.

The downside is context switching. Some weeks I had different codebases open and had to remember which architecture decisions belonged to which project. I’ve gotten better at it. My secret: extensive notes. Not fancy systems. Just a text file per project with decisions, open questions, and things that confused me. Future me always appreciates past me’s notes.

Go

I kept contributing to open-source Go projects. Nothing dramatic: bug fixes, documentation improvements, the kind of work that keeps an open-source project healthy. Go remains my language of choice for production systems. It’s boring in the best way. The code I write in Go today looks like the code I wrote three years ago, and I like it that way.

AI tooling in Go is still immature compared to Python. I find myself writing Go wrappers around Python services more than I’d like. But I’d rather have a reliable Go service calling a Python sidecar than a Python monolith that I have to babysit.

What Stayed Hard

Evaluation. I wrote about it multiple times this year because it remained the hardest unsolved problem in AI engineering. Everyone agrees it matters. Nobody has a great solution for multi-step workflows. I got better at building lightweight eval suites, but they’re still more art than science.

Trust. One confidently wrong answer can undo weeks of user adoption. I saw this happen at two different companies this year. The AI feature was great 95% of the time and catastrophically wrong 5% of the time, and users only remembered the 5%.

Cost management. Token-based pricing sounds simple until you multiply it by production volume and realize your prompt changes have budget implications. I now review prompt changes like I review infrastructure changes, with a cost estimate attached.

Looking at 2024

I don’t do predictions. But I know what I’m going to focus on: making AI systems more reliable and more auditable. The hype cycle will do what hype cycles do. The engineering work of making these systems trustworthy is the real job, and it’s the job I want to be doing.

2023 was the year AI became real. 2024 will be the year we find out if it can stay real.

Happy holidays. Go take a break. The codebase will be there when you get back.