TLDR: Harness has launched its AI-powered DevSecOps agents, a unified platform designed to automate and streamline the post-coding stages of software delivery. This initiative directly addresses the growing ‘bottleneck’ created by rapid AI code generation, focusing on improving testing, security, deployment, and optimization to enhance speed, reliability, and efficiency.
In a significant move to revolutionize software development, Harness has introduced its new AI-powered DevSecOps agents, aiming to tackle the critical bottlenecks emerging in the software delivery lifecycle. While artificial intelligence has dramatically accelerated code generation, enabling developers to produce thousands of lines of code in minutes, the subsequent stages—testing, securing, deploying, and optimizing—have become the new pain points, consuming up to 80% of engineering time.
According to Harness CEO Jyoti Bansal, the real challenge lies ‘after you hit “commit.”’ He highlighted the extensive ‘toil’ involved in managing CI/CD pipelines, test suites, security vulnerabilities, and escalating cloud costs. Bansal noted, “You could spend 35 to 45 hours a week just managing and maintaining a CI/CD pipeline,” adding, “Your build failed? That’s another hour. Cloud costs spiking? There goes your afternoon.” Data from Harness indicates that nearly 80% of software failures occur post-coding, with CI/CD pipelines being the primary culprit. The 2024 DORA report further corroborates this, showing a decline in software delivery stability and throughput despite the proliferation of AI tools.
To counter this, Harness has developed Harness AI, a unified platform that automates every step after code generation. This platform is built around a ‘Software Delivery Knowledge Graph,’ an intelligent layer that continuously updates and connects data across the entire Software Development Lifecycle (SDLC), including builds, tests, deployments, incidents, infrastructure changes, and cloud spend. This knowledge graph provides the context necessary for Harness AI to make informed decisions, ensuring automation is precise, secure, and aligned with organizational policies.
The platform leverages specialized AI agents—including DevOps, SRE, Release, AppSec, Test, and FinOps agents—which operate seamlessly, often through a simple chat interface. These agents are designed to generate enterprise-grade pipelines, intelligently create and maintain tests and chaos experiments, detect and remediate vulnerabilities in real-time, and optimize cloud spend by providing actionable insights and recommendations.
Early results from enterprise customers demonstrate significant improvements: test cycle times have been reduced by 80%, downtime slashed by over 50%, test maintenance effort cut by 70%, and pipeline debugging time decreased by 50%. Bansal remarked on the positive feedback, stating, “People are saying nothing like this exists. Everyone’s struggling with DevOps and security and all these things that eat up 70% of engineering time, and there’s no good AI solution for it.”
The AI layer of Harness AI has been under development for approximately 30 months, with collaborations extending to leading AI providers such as Anthropic, OpenAI, Google Gemini, AWS, and developer tools like Windsurf and Cursor. Bansal emphasized that as AI makes code generation easier, the focus must shift to ensuring the rest of the software delivery pipeline can keep pace. He stated in a blog post that Harness AI is “the foundation for the next generation of software delivery — a generation that delivers software not just for today’s applications but also for AI-powered applications with the same speed, reliability and security.”
Also Read:
- Pulumi Unveils Neo: An AI Agent Revolutionizing Cloud Infrastructure Management
- Edward Jones Transforms Software Engineering with Generative AI, Boosting Efficiency and Innovation
Harness AI is now generally available to all Harness customers, with a broader showcase anticipated at the company’s Unscripted conference in September.
