What I Actually Use AI For Day-to-Day

I use AI every day. Not in a vague, exploratory way — in a concrete, this-is-now-part-of-how-I-work way. I’m also leading the development AI-first approach at FINN Partners, which means I’m thinking about how it fits into team workflows, not just my own. Here’s what that actually looks like.

Claude Code

This is where I spend most of my time. Claude Code is a CLI tool from Anthropic that runs in your terminal alongside your editor. It can read your codebase, write and edit files, run commands, and reason through problems with context.

I use it for the kind of work that used to require deep focus but not necessarily deep thinking — boilerplate, repetitive refactoring, wiring things together, scaffolding new projects with our standard stack in place from the start. It handles the mechanical parts, and I handle the decisions. That split works well once you get used to it.

It’s also useful when moving across a large codebase quickly. Having something that can read context and surface relevant detail speeds up that process significantly.

Building with the Claude API

Beyond using Claude as a tool, I’ve been building with it directly via the Claude API — internal automations, workflow tools, small experiments. The API is straightforward enough that you can go from idea to working prototype quickly, which is where a lot of the value is. Getting something functional in front of people fast, then iterating.

Replit for Internal Tools

Replit is quick to spin up and easy to share with non-technical stakeholders. I use it for internal tools that sit outside the main codebase.

Code Review

AI has changed how I approach code review. Before looking at a PR in detail, I’ll often run it through a model to get an initial read — obvious issues, potential edge cases, anything worth flagging. It doesn’t replace the actual review, but it means I’m going into it with more context and a clearer sense of where to focus.

It’s also useful in the other direction: before submitting my own work, having something review it independently gives a perspective that’s hard to get when you’ve been in the same codebase all day.

The next step is automating this at the team level — running AI reviews on every PR as part of the pipeline, so there’s always a first pass in place before a human picks it up. Still building that out, but the direction is clear.

Testing

Writing tests has historically been one of those tasks that everyone agrees is important and nobody enjoys as much as they should. AI is genuinely good at this — generating test cases, surfacing edge cases worth covering, writing the repetitive parts. I still review everything, but the starting point is much better.

The Job Anxiety Thing

People ask me about this fairly regularly, especially junior developers. I understand why — the coverage around AI and jobs is hard to ignore.

What AI can’t do is understand the business problem, make the right tradeoffs, know when something is technically correct but wrong for the context. That’s still the work.

Most developers have access to these tools now. The ones being left behind are those treating it as a passing trend. That gap is only going to widen.

Where This Is Headed

Few have figured out how to use them consistently as a team — shared standards, review processes, where AI fits and where it doesn’t. That’s what I’m focused on at FINN Partners.

If you’re not using AI in your day-to-day yet, the barrier to starting is lower than you probably think. Pick one workflow, try it for a week, and see what changes.

If you’re working through any of this — tooling, process, team adoption — feel free to reach out on LinkedIn.