Harness DevOps tools add database, QA automation, AI agents
With a series of updates this week, Harness claims unique support for database DevOps, as well as pipeline generation and UI test automation using AI agents.
Harness previewed three new modules for its DevOps tool set, a group of generative AI agents that automatically create DevOps pipelines and QA tests, and its own software artifact registry with supply chain security baked in.
Harness expects the series of product updates to become generally available in early 2025. The company, which began with a focus on continuous delivery in 2018, markets a set of modules that can be used together as a single DevOps platform or swapped out with third-party tools.
The forthcoming modules unveiled by Harness include the following:
Cloud Development Environments. Similar to Docker Build Cloud and GitHub Codespaces, this module replaces developer laptops with a cloud-based VM prepared by platform engineers with production-like dependencies built in.
Artifact Registry. A software artifact registry is used to store and manage the products of the process by which raw code is built into deployable components. There are other well-established alternatives, including JFrog Artifactory, Sonatype Nexus Repository and offerings by all three major public cloud providers. Harness co-founder and CEO Jyoti Bansal acknowledged this but said his company's module has been designed with caching support that makes its scalability and performance stronger than alternatives, as well as a chain of custody and digital signing for artifacts that follows them through deployments with Harness's CI/CD tools.
Database DevOps. Here, Bansal also claimed Harness will expand on database DevOps tools such as Liquibase and Redgate's Flyway that focus on aligning database change management with the CI/CD process. The Harness module will create CI/CD pipelines for database changes synced with application code changes, with the company's automated rollback features included, he said.
Database rollback entices DevOps director
The Database DevOps module caught the attention of one Harness customer who was briefed by the vendor on its planned updates.
Jignesh Patel
"Database changes are seen as an easy one-way door, [but] rollback is always a struggle," said Jignesh Patel, director of cloud and DevOps at Morningstar, a financial services company headquartered in Chicago. "Bringing it into the broader platform means testing [will be] better integrated, as well as the execution [of updates]."
Morningstar switched from CloudBees and Jenkins to Harness two years ago in part because Patel had liked working with Harness at United Airlines before he joined the company. The Harness method of using variable expressions to capture pipeline configurations also cut down more than half of the CI/CD pipelines Morningstar had been managing with Jenkins. Built-in but customizable integrations under Harness appealed to developers who'd had to build their own plugins between third-party tools and Jenkins.
"One of our teams said they removed 2,000 to 3,000 lines of code from their applications" by removing that custom integration, according to Patel. Morningstar's DBAs will take some convincing to move to Harness Database DevOps, but he said he hoped that experience for other teams would convince them to try it out.
Patel said he'd also consider the Harness Artifact Registry, even though his company is currently using a competitor.
"We may go to Harness [Artifact Registry] just because it probably integrates better with their platform over using JFrog as an integration point," he said.
Other Harness customers are also likely to be interested in a built-in artifact registry, said Andrew Cornwall, an analyst at Forrester Research.
"It's too early to say how they compare, but in many cases the customers who buy all-in-one DevOps platforms like Harness aren't particular about their artifact registry -- as long as they have one, they're happy," Cornwall said. "I don't know if the same is true of their cloud development environment -- developers will want to play with it and verify that it suits their needs before adopting it."
Harness looks past coding assistant with AI agents
This week's previews also included generative AI agents, most of which will be added to existing modules, and a standalone AI QA Assistant that will be sold separately. The tools previewed this week will be part of a library of generative AI agents built by Harness based on foundational large language models from Google's Gemini line, OpenAI and open source projects such as StarCoder.
AI Code Assistant is similar to existing coding assistants such as GitHub Copilot, Bansal acknowledged, other than what he claimed will be a better developer experience. This feature will be available free for Harness customers.
AI DevOps Assistant marks a foray into generative AI agent automation for Harness, to automatically generate CI/CD pipeline code, remediate failures and suggest optimizations. This assistant will be bundled with Harness CI and CD modules.
AI Productivity Insights adds features to the Harness Software Engineering Insights module based on its acquisition of Propelo in 2023 that collects velocity and code quality DORA metrics for developers using AI assistants to compare against those who aren't within a given organization.
AI QA Assistant is a standalone product and will represent a breakthrough in UI test automation, according to Bansal. It will generate and update code to test mobile and web user interfaces for applications on the fly in response to natural language requests, a departure from typical manual test scriptwriting by human QA engineers.
Cornwall said he's impressed with the amount of new features Harness has been able to generate in a relatively short period of time over the last year, especially the development of its AI agents.
The DevOps Assistant … will improve the lives of DevOps engineers significantly. They'll offload a lot of the tedious parts of managing pipelines.
Andrew CornwallAnalyst, Forrester Research
"The most forward-thinking is the DevOps Assistant, which will improve the lives of DevOps engineers significantly. They'll offload a lot of the tedious parts of managing pipelines," Cornwall said. "The QA Assistant lets Harness compete with testing platforms that target subject-matter expert and business testers as well as professional testers and developers."
Morningstar's Patel said he's keeping an open mind about the new AI agents but is unsure whether Harness can surpass vendors that have offered generative AI assistants for a longer period, especially GitHub Copilot.
"They're not quite there yet," Patel said of Harness's catch-up to GitHub Copilot. "The recommendations aren't as in depth. … To be frank, they just don't have the same code base at hand to train models."
Still, the AI QA Assistant is worth exploring further, Patel said.
"I am curious to take that for a spin, because you can know about broken links or missed tests a lot faster than an individual can [go through] all the scenarios," he said.
Beth Pariseau, senior news writer for TechTarget Editorial, is an award-winning veteran of IT journalism covering DevOps. Have a tip? Email her or reach out @PariseauTT.
Jignesh Patel
