This is the first time I’ve onboarded somewhere with modern LLMs already on tap. The experience has been better.
I joined Buffer, an async-first remote company that builds tools for creators and businesses to manage their social media. The team’s been great at providing context and guidance, but some things in engineering you only learn by doing. AI tools have made starting that part easier than I expected, especially for building context.
These tools have been most useful for research. When I run into an unfamiliar pattern or an architecture decision, I ask questions until I understand why the code is the way it is. I still apply my own judgment on what to do next. AI just gets me to that point faster.
The GitHub CLI and the Linear and Notion agents have been useful here. Instead of spending hours hunting through repos and docs, I get pointed in roughly the right direction quickly, then dig in myself.
It extends past the codebase too. When I need to understand how we handle things internally, retrieval, storage, validation, that sort of thing, I point an agent at our docs and have it research how it actually works. Then I can propose something that fits our existing patterns instead of guessing.
Local setup issues are still a thing, of course. AI has been good at parsing cryptic error messages and suggesting fixes. I don’t have to wait on someone to be around. In an async environment, immediate help isn’t always there.
Plan mode has been useful too. Before committing to a solution, I explore a few approaches and weigh the tradeoffs. I evaluate what the model suggests with whatever engineering experience I’ve built up, and decide from there.
It’s tempting to work in a silo when you can unblock yourself more easily. But that misses the point. I still reach out, ask questions, and pair on the harder problems. I just have better context and sharper questions.
Past onboardings took longer. This one’s been faster.