A brief summary of my experience with Claude Code
I've been using Claude Code professionally for the past 2 months, with limited agent use prior to that (via Windsurf). Overall, I've seen a roughly 30% boost in productivity in full-stack development, with significant spikes in particular types of work.
Where Claude Code has excelled:
Building well-defined features built upon existing conventions (10x+ boost)
Performing similar mid-level changes across multiple files (10x+ boost)
Executing large refactors or architecture changes (10x+ boost)
Performing analysis of existing codebases to help build my personal understanding (10x+ boost)
Heavy pattern-matching tasks, which interestingly includes configuring complex or nested UI layouts.
Where Claude Code has floundered or wasted time:
Anything involving temporal glitches in UI or logic. The necessary feedback loop just can't yet be accomplished with normal tooling.
Fixing state issues in general. Again, the feedback loop is too immature for CC to even understand what to fix unless your tooling or descriptive ability is stellar.
Solving classes of smallish problems that require a lot of trial-and-error, aren't covered by automated tests, or require a steady flow of subjective feedback. Sometimes it's just not worth setting up the context for CC to succeed.
Adhering to unusual or poorly-documented coding/architecture conventions. It's going to fight you the whole way, because it's been trained on conventional approaches.
Productivity hacks:
Run multiple agents on multiple tasks for true multitasking. These agents are automated, meaning you can literally have multiple streams of work performed in parallel. CC may not beat a single engineer for many types of tasks, but it can literally do multiple things at once. This is where a lot of real productivity-boosting potential comes into play.
This doesn't come without a real human challenge: monitoring multiple projects and maintaining your own human mental context for each. I've found myself more mentally exhausted at the end of the day.
Invest in good context documents as early as possible, and don't hesitate to ask CC to insert new info and insights in its documents as you go. This is how you can help CC "learn" from its mistakes: document the right way and the wrong way when a mistake occurs.
If possible, run Claude Code within a dev container with just the tools it needs. This allows you to run CC with full permissions on tool use without concern that it will perform dangerous actions. Otherwise, you may find yourself constantly unblocking CC as it requests permission to use the tools it needs.
Personal Context:
I'm a 16-YoE fullstack engineer at a startup, working with React/Remix, native iOS (Swift + UIKit), native Android (Kotlin + Jetpack Compose), backends in TypeScript/Node, and lots of GraphQL and Postgres. I also use Elixir (with CC) and React Native for personal projects. I've had success with Claude Code in all of these codebases and languages.