Harness, the software delivery platform company, has unveiled its AI assistant, AIDA (AI Development Assistant), which the company calls “a game-changer for the entire software development life cycle (SDLC)”.
“This innovative approach to AI for software delivery stands in contrast to traditional AI applications in the field that primarily focus on generating and assisting with code development. This comprehensive AI solution is available free of charge to all Harness customers and will be seamlessly integrated across all Harness platform workflows and capabilities, including continuous integration (CI), continuous deployment (CD), cloud cost management, and feature flags,” says the company.
"Harness's goal has always been to break down barriers in software delivery. With the launch of AIDA, we're not just talking about a future where AI significantly enhances the SDLC, we're making it a reality," says Jyoti Bansal, chief executive officer and co-founder of Harness. "I firmly believe this is just the beginning, and I'm incredibly proud of our team's relentless efforts to drive this change. We're committed to harnessing the potential of generative AI to address real-world developer challenges across the entire lifecycle of code."
Harness forecasts a 30-50% boost in software engineering teams' productivity with generative AI. With AIDA, Harness is expanding the influence of AI beyond coding to all other facets of the SDLC, which includes building and testing code, ensuring security, ensuring reliability, deploying changes, and optimising costs.
“AIDA debuts today with three groundbreaking features designed to alleviate key pain points in the software development lifecycle,” adds Bansal.
“First is the assisted resolution of builds and deployment failures. AIDA can analyse log files and correlate error messages with known issues. This feature enables developers to troubleshoot and resolve deployment failures quickly, saving them from sifting through millions of log lines. AIDA also suggests fixes and predicts potential errors in the code even before the build is initiated. This feature is designed to work across Harness’s CI and CD offerings.”
“Second is the automated security vulnerability fixing. AIDA can automatically identify security vulnerabilities and generate code fixes. Trained on all publicly known common vulnerabilities and exposures (CVEs) and common weakness enumerations (CWEs), AIDA significantly accelerates the vulnerability remediation process, reducing developer effort by 50-75%. This feature is designed to integrate seamlessly with the Harness Security Testing Orchestration module.”
“Third is the managed cloud assets using natural language. AIDA simplifies the process of creating and maintaining policies to manage cloud costs. Using natural language, users can define policies for governing cloud assets and costs, automating a typically complex process.”
Additional AI features will be incorporated in AIDA over three to six months, including automated code reviews, AI-assisted CI/CD pipeline authoring, and AI-assisted authoring of chaos engineering experiments.
Adopters of AI in developer tooling, especially in enterprises, are concerned about data privacy, security, and legal issues regarding code copyright infringement.
“AIDA is designed to address these concerns from the beginning. The Harness AI stack, composed of Harness custom LLMs, training data, vector databases, etc., are all specifically selected and designed, so all source code used in training have permissive licenses. This allows users to freely incorporate the results from the Harness AI stack without being concerned about copyleft requirements or the risks of using non-open sourced, proprietary code. There is no proprietary or copyleft code used for training the Harness AI stack. Additionally, any code and data shared by customers with AIDA remains fully secure and always owned by the customers,” explains Bansal.
“Harness, with its rich history of pioneering AI/ML in software delivery, emerged as a market disruptor five years ago using AI to predict failures from code changes. We pioneered the use of AI in its continuous delivery-as-a-service platform, the first of its kind, which launched in 2017. Building upon Harness's legacy of AI for DevOps, AIDA is now redefining and reimagining the use of generative AI for software delivery.”