Adronite Unveils AI Coding Tool Based on Embedded Context Engine

Adronite has unveiled an AI coding tool called Codistry, which harnesses a context engine known as the Adronite Context Engine (ACE). According to Dr. William T. Colleran, CEO of Adronite, ACE understands relationships within a codebase to generate better code at a lower total cost due to more efficient token consumption. Prior to code generation, ACE maps the software architecture, dependencies, vulnerabilities, and relationships within the codebase. This enables Codistry to better predict the impact of generated code on the existing codebase.

The tool is designed to work with self-hosted open-weight AI models, eliminating the need to share large portions of a codebase with AI models and reducing token consumption. Adronite claims its internal benchmarking shows that Codistry completes comparable development tasks using roughly half the tokens, resulting in a 48% cost reduction. For instance, on the PocketBase open-source codebase, Codistry reduced per-task costs from $2.12 to $1.10.

Originally, ACE was designed for a documentation tool, but its potential in AI coding tools became evident. Developers can express their intent more easily without needing a deep understanding of the underlying codebase to generate usable code in a production environment. Mitch Ashley, vice president and practice lead for software lifecycle engineering at The Futurum Group, notes that context is crucial as developers now spend more time explaining gaps in context than reviewing AI model output.

Adronite's context engine maps dependencies before code generation, benefiting the code review process.

While AI coding tools are widely adopted, the quality of generated code varies, leading to more post-deployment issues for DevOps teams. Regardless, AI models are becoming interchangeable commodities, and a tool like Codistry significantly reduces the cost of switching between models. Developers may even test multiple AI models for the same task to determine the best output.

Ultimately, the challenge lies in ensuring the created code is of higher value than any potential issues it might cause downstream in the DevOps pipeline.