ARTHUR LABS

**ArthurLabs — Think. Build. Verify.** ArthurLabs is an AI ecosystem built around three connected ideas: understanding what we want to create, turning those ideas into working software, and verifying that the result actually works. Today, AI tools can generate code incredibly quickly, but the development process is still fragmented. Ideas get lost across conversations, coding agents can lose context or report tasks as complete without sufficient verification, and reviewing AI-generated code often means relying on another AI's opinion. We're building ArthurLabs to address these problems through three complementary projects. **Arthur Studio — Think** Arthur Studio helps users turn scattered thoughts into connected, evolving knowledge. Instead of simply storing notes, it structures ideas, identifies meaningful relationships and contradictions, and preserves how thoughts change over time through a Cognitive Graph and Thought Lineage. **Nexus — Build** Nexus focuses on reliable AI-assisted software engineering. We have developed two distinct approaches: Nexus DSH, which extends DeepSeek's agent harness with structured execution and verification workflows, and Nexus Native, our original Rust-based agent harness with its own execution controller, verification, and recovery mechanisms. We're also exploring Nexus Nova, a planned extension of Native designed to adapt agent coordination to different task complexities. **Sentry — Verify** Sentry provides independent AI-powered code review and quality assurance. It uses multiple independent analysis paths to examine the same codebase, collect findings, and connect them to supporting evidence and test results. Its goal is to make software review more transparent and reliable rather than simply trusting a single AI response. **How They Work Together** The three projects address different stages of the same development process. Arthur Studio helps organize and understand ideas. Nexus provides the engineering systems needed to turn those ideas into software. Sentry independently challenges and reviews the resulting code. They are currently distinct systems rather than one fully automated pipeline, but together they represent our vision for a more connected, accountable, and trustworthy approach to AI-assisted development. **ArthurLabs is about going beyond code generation — from thought, to engineering, to verification.**