---
title: "How Netflix Made AI Coding Agents More Efficient with JetBrains IDE Intelligence"
description: "Netflix integrated JetBrains’ headless code-intelligence engine into coding agents Claude Code and Codex to address agents’ struggles navigating Java/Kotlin code across compiled JAR boundaries. The engine provides deterministic indexing of project source files and libraries, and Netflix built a daemon and CLI to make the LSP callable by agents, keeping the index warm across repeated calls. In benchmarks on two real codebases—an internal Spring Boot service and the open-source DGS Framework—agents using JetBrains intelligence were typically 20-30% more efficient in tokens and time, with up to 65% improvement for stronger models like Opus 5 at high effort. Correctness held steady or improved across all configurations, and the gains were consistent across every model and agent tested. Netflix plans to extend the same benchmark methodology to Python, TypeScript, and Go, as coding agents for those languages face similar problems."
---

# How Netflix Made AI Coding Agents More Efficient with JetBrains IDE Intelligence

[Jetbrains](https://yomu.fyi/company/jetbrains) · Anastasia Kazakova · Oct 15, 2026

**Type:** Benchmark

## Summary

Netflix integrated JetBrains’ headless code-intelligence engine into coding agents Claude Code and Codex to address agents’ struggles navigating Java/Kotlin code across compiled JAR boundaries. The engine provides deterministic indexing of project source files and libraries, and Netflix built a daemon and CLI to make the LSP callable by agents, keeping the index warm across repeated calls. In benchmarks on two real codebases—an internal Spring Boot service and the open-source DGS Framework—agents using JetBrains intelligence were typically 20-30% more efficient in tokens and time, with up to 65% improvement for stronger models like Opus 5 at high effort. Correctness held steady or improved across all configurations, and the gains were consistent across every model and agent tested. Netflix plans to extend the same benchmark methodology to Python, TypeScript, and Go, as coding agents for those languages face similar problems.

## Context

Netflix’s coding agents struggled to navigate Java/Kotlin code, especially across compiled library (JAR) boundaries, because agents had to unzip JARs and decompile class files. JetBrains’ code-intelligence engine, which powers IntelliJ IDEA, provides deterministic indexing of source and libraries. Netflix integrated a headless version of this engine into agents via an LSP, building a daemon and CLI to manage the index lifecycle.

## Approach / What changed

Netflix built a daemon and CLI in Java, distributed as a GraalVM executable, to expose JetBrains’ headless code-intelligence engine via LSP to coding agents. The daemon keeps a warm index in memory across calls, avoiding cold-start overhead. The CLI provides a stateless interface for each tool call, routing through the daemon’s session manager.

## Takeaways

- JetBrains IDE intelligence improved coding agent efficiency by 20-30% typically, up to 65% for stronger models, across all tested agents and tasks.
- The key architectural component is a daemon that keeps the project index warm, avoiding multi-second cold starts on repeated LSP calls.
- Correctness held steady or improved in every benchmark run, meaning the efficiency gains did not come at the cost of task completion quality.

## FAQ

### How does the daemon avoid cold-start overhead?

The daemon's session manager routes each call by project root to a matching server process that holds the project's index in memory, so subsequent calls skip indexing and go straight to a warm server.

### What was the best-case efficiency gain?

The best case was about 65% improvement, achieved with the strongest model (Opus 5) at high effort on code navigation tasks.

### Did using JetBrains intelligence affect correctness?

Correctness held steady or improved in every run across all models and agents, so the benefit was purely in efficiency without sacrificing task completion.

**Tags:** [Coding Agents](https://yomu.fyi/topic/coding-agents), [Java](https://yomu.fyi/topic/java), [JetBrains](https://yomu.fyi/topic/jetbrains), [LSP](https://yomu.fyi/topic/lsp), [Netflix](https://yomu.fyi/topic/netflix)

- Source: [Jetbrains](https://blog.jetbrains.com/ai/2026/10/how-netflix-made-ai-coding-agents-more-efficient)
- Source URL: https://blog.jetbrains.com/ai/2026/10/how-netflix-made-ai-coding-agents-more-efficient
- Ingested by Yomu: 2026-10-15T12:00:18.672Z

[Read original post](https://blog.jetbrains.com/ai/2026/10/how-netflix-made-ai-coding-agents-more-efficient)