I use LLMs as a staff engineer in 2026
In 2026, staff engineers increasingly rely on advanced LLM agents to generate complete pull requests and diagnose most bugs, marking a shift from earlier uses like autocomplete and one-off code queries. Agents now operate effectively across multiple repositories and require only light human review, though engineers still oversee and validate outputs. Human expertise remains critical, especially in complex debugging and high-level communication, where LLMs struggle with concision and contextual understanding.
- ▪LLMs now generate entire pull requests in familiar codebases, with engineers typically making only one editing pass before merging.
- ▪Agents diagnose about 80% of bugs independently, especially when given access across multiple repositories.
- ▪Engineers still write their own PR descriptions to better convey the core idea and signal personal review.
- ▪Current agents recover from errors autonomously and operate too quickly for real-time human intervention.
- ▪Human insight remains essential in narrowing down complex bugs, even when an agent ultimately identifies the issue.
Opening excerpt (first ~120 words) tap to expand
How I use LLMs as a staff engineer in 2026A bit over a year ago I wrote How I use LLMs as a staff engineer. Here’s a brief summary of what I used AI for last year: Smart autocomplete with Copilot Short tactical changes in areas I don’t know well (always reviewed by a SME) Writing lots of use-once-and-throwaway research code Asking lots of questions to learn about new topics (e.g. the Unity game engine) Last-resort bugfixes, just in case it can figure it out immediately Big-picture proofreading for long-form English communication Here are some tasks I explicitly didn’t use AI for last year: Writing whole PRs for me in areas I’m familiar with Writing ADRs or other technical communications Research in large codebases and finding out how things are done February 2025 was a long time ago.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Seangoedecke.
