About

Brainlet exists because code assistants still do not understand projects.

Brainlet is a project intelligence engine for LLM-powered software development. It builds a computed understanding layer from a codebase so models can work with architecture, conventions, dependencies, data flow, and impact paths instead of guessing from retrieved files.

Local-first by default

Brainlet is designed so indexing and project intelligence can run where the repository already lives: a developer machine or company-controlled server.

Context before generation

The core belief is simple: an LLM should receive project knowledge before it writes, reviews, or explains code.

Small engine, clear boundary

Brainlet does not replace the model. It builds the project understanding layer the model can query.

Founder

Built by Younes Rezzouki.

Younes Rezzouki is the founder and CEO of Brainlet. He has 15 years of software engineering experience across five countries, including engineering leadership in fintech.

Brainlet started as a practical response to a daily problem: AI coding tools could retrieve files, but they still missed the project-level reasoning that senior engineers use during review and implementation.

Founder LinkedIn

Origin

Started as a local tool for making AI coding assistants less blind to architecture and project conventions.

Engine

Built around a Rust-native analysis pipeline for code, configuration, relationships, and project-level signals.

Product

First public product focus: project-aware PR review, with public benchmark methodology planned for June 2026.