• About Us
  • Contact us
  • DMCA
  • Home
  • Privacy Policy
  • Subscribe to our Newsletter
Wednesday, November 18, 2026
No Result
View All Result
NEWSLETTER
The San Francisco Tribune
  • Home
  • Art
  • Business
  • Entertainment
  • Sports
  • Food
  • Magazine
  • Podcasts
  • Politics
  • Tech
  • Wellness
  • Home
  • Art
  • Business
  • Entertainment
  • Sports
  • Food
  • Magazine
  • Podcasts
  • Politics
  • Tech
  • Wellness
No Result
View All Result
The San Francisco Tribune
No Result
View All Result
Home Tech

OpenHands Wants to Turn AI Coding Agents Into Software Factories, Starting With Cybersecurity

by Editorial
November 17, 2026
in Tech
0
OpenHands Wants to Turn AI Coding Agents Into Software Factories, Starting With Cybersecurity
Share on FacebookShare on Twitter

AI is accelerating software development and cyber threats, but fixing vulnerabilities remains a stubbornly manual process. OpenHands believes AI-powered software factories could help close the gap.

Artificial intelligence is transforming software development, enabling programmers to write code, debug applications, and complete engineering tasks faster than ever before. But as AI accelerates the pace of software development and cybersecurity threats, the process of fixing security vulnerabilities is struggling to keep up.

According to Verizon’s 2026 Data Breach Investigations Report, the median time organizations took to fully remediate known exploited vulnerabilities increased from 32 days to 43 days over the previous year. Even more concerning, only 26% of known exploited vulnerabilities in the study were fully remediated, down from 38% a year earlier.

The volume of newly discovered vulnerabilities is also putting pressure on engineering teams.

“More than 17,000 CVEs were published in October, four times what we saw in January,” said Robert Brennan, CEO and co-founder of OpenHands. “Engineering teams are drowning trying to keep production systems secure, and it’s becoming clear that human defenders alone can’t keep pace with agentic attackers.”

The challenge is that identifying vulnerabilities is becoming increasingly automated, but fixing them still requires substantial engineering work.

When a vulnerability is discovered, developers must investigate the affected code, determine the appropriate fix, implement changes, run tests, and coordinate reviews before the issue can be resolved. Multiply that effort across hundreds or thousands of vulnerabilities, and even well-resourced engineering organizations can struggle to keep pace.

OpenHands, the company behind an open-source platform for AI coding agents, believes the solution lies in a new approach to software engineering: the AI-powered software factory.

Rather than simply helping individual developers write code faster, OpenHands wants to enable organizations to automate entire engineering workflows, allowing AI agents to continuously investigate problems, implement solutions, test changes, and prepare their work for human review.

And cybersecurity vulnerability remediation could be one of the most compelling applications of the technology.

From AI Coding Assistants to Software Factories

The first wave of AI coding tools focused largely on individual developer productivity. Products such as GitHub Copilot and Cursor demonstrated how large language models could help programmers write code, understand unfamiliar applications, and accelerate routine development tasks.

More advanced AI coding agents have since emerged, capable of independently navigating repositories, modifying files, executing commands, and running tests.

But OpenHands sees an opportunity to take the technology a step further.

Instead of requiring a developer to initiate each task, a software factory uses AI agents as part of a coordinated system that can continuously execute engineering work. These workflows can be triggered automatically by events such as a newly discovered vulnerability, a failed build, or an issue submitted through a development platform.

The distinction is significant. Rather than simply making an individual programmer more productive, the software factory automates repeatable processes that might otherwise consume hours of engineering time.

Consider a security scanner that identifies a critical vulnerability in an application. Traditionally, that finding would enter a backlog, where an engineer would eventually investigate the issue, develop a fix, and submit the changes for review.

In an AI-powered software factory, the security finding could instead trigger an automated workflow. An AI agent would examine the vulnerability, inspect the relevant code, determine a potential remediation, implement the changes, and run tests to validate the proposed fix. If successful, the system could generate a pull request for an engineer to review and approve.

The objective isn’t necessarily to produce more code. It’s to complete more of the engineering work required to maintain, improve, and secure existing software.

Why Cybersecurity Is a Natural Starting Point

Vulnerability remediation offers a particularly attractive use case for autonomous software engineering because it combines a growing operational burden with relatively well-defined engineering tasks.

Organizations already rely on automated security tools to discover vulnerabilities in applications, dependencies, and infrastructure. These tools generate findings that must be prioritized and addressed, often across large numbers of repositories and development teams.

But the ability to discover vulnerabilities has outpaced the capacity of many organizations to remediate them.

This creates an uncomfortable imbalance: security teams can identify potential weaknesses at machine speed, while engineering teams are still responsible for much of the remediation work at human speed.

OpenHands believes AI agents can help address that imbalance by automating the repetitive work between vulnerability discovery and a proposed, tested fix.

In March 2026, the company introduced Vulnerability Fixer, an open-source application that demonstrates how AI agents can automate much of the remediation process.

The application uses a four-stage workflow. First, it scans a GitHub repository using Trivy or processes vulnerability reports from security tools such as Snyk, Veracode, and Checkmarx. It then analyzes the findings and identifies vulnerabilities to address.

From there, AI agents investigate the relevant code, generate potential fixes, and run tests to validate the changes. Finally, the system creates GitHub pull requests containing the proposed remediations, allowing developers to review the changes before they are merged.

The application also supports multiple agents working simultaneously on different vulnerabilities, with a dashboard for monitoring their progress and reviewing execution logs.

Brennan envisions a future in which these systems go beyond preparing fixes for review and can resolve certain vulnerabilities without requiring engineers to intervene.

“Our goal is a world where a CVE gets announced at 2 a.m. on Saturday and patched in production by 3 a.m., without an engineer needing to wake up,” Brennan said.

It’s an ambitious vision, particularly given the risks associated with automatically modifying production software. Not every vulnerability can be resolved through a straightforward code change. Some require architectural decisions, changes to business logic, or an understanding of operational risks that automated systems may struggle to evaluate.

Even when an AI agent produces a working patch, passing automated tests doesn’t necessarily establish that the underlying security problem has been fully resolved.

Brennan acknowledges that achieving greater autonomy will require more than simply improving the capabilities of AI agents.

“Achieving that level of autonomy will require thoughtful policies, strong guardrails, and trust in the systems doing the work,” he said.

For now, OpenHands’ approach emphasizes automating the repetitive investigation and implementation work while allowing developers to evaluate proposed changes and determine whether they should be deployed.

Building the Infrastructure for Autonomous Engineering

Of course, automating a single vulnerability fix is very different from running hundreds of AI agents across an enterprise.

Organizations must consider how agents access source code, which tools and credentials they can use, where they execute their work, and how their actions are monitored. They also need mechanisms for controlling costs, enforcing security policies, and ensuring that automated changes do not introduce new problems.

These challenges are central to OpenHands’ strategy.

The company provides an open-source foundation for building AI coding agents, along with infrastructure for executing and managing them across development environments.

Its automation capabilities allow organizations to establish workflows that run on schedules or respond to events from development systems. Meanwhile, the OpenHands Agent Control Plane provides centralized capabilities for managing agent execution, access, usage, and organizational oversight.

That combination is intended to move AI agents beyond isolated experiments and into repeatable engineering workflows.

OpenHands also emphasizes flexibility as a differentiator. Rather than requiring customers to standardize on a particular language model, the platform supports multiple models, allowing organizations to choose based on performance, cost, and infrastructure requirements.

Its open-source foundation gives developers the ability to inspect, customize, and extend the underlying technology. Organizations can also run agents in their own environments, an important consideration for enterprises concerned about where their source code is processed.

These capabilities become especially important when AI agents are working with sensitive source code and security vulnerabilities.

Giving autonomous systems access to development environments introduces risks of its own, including unintended code changes, excessive permissions, and exposure to malicious instructions. Organizations need safeguards that limit what agents can access and do, while maintaining visibility into their actions.

For OpenHands, the challenge is not simply making AI agents more capable. It’s providing the infrastructure that allows organizations to operate those agents reliably, securely, and at scale.

A Different Future for Software Engineering

Although vulnerability remediation provides a practical starting point, OpenHands sees software factories as a much broader opportunity.

The same approach could be applied to dependency upgrades, automated code reviews, failed build investigations, software testing, and large-scale codebase migrations.

These are tasks that consume substantial engineering resources but often follow repeatable patterns, making them candidates for automation.

If the technology matures as OpenHands envisions, software development could increasingly shift from engineers manually executing individual tasks to engineers defining, supervising, and improving automated workflows.

Instead of assigning a developer to investigate every failed build or security alert, organizations could establish systems that continuously perform the initial work and escalate decisions requiring human judgment.

That transition is unlikely to happen overnight. Autonomous agents still face challenges involving reliability, accuracy, context, and the complexity of real-world software systems. Organizations will also need to determine which workflows can be safely automated and how to measure whether the resulting work actually improves engineering productivity.

Nevertheless, the direction is becoming clearer: the next phase of AI-powered software development may be less about how quickly an individual developer can write code and more about how much engineering work an organization can complete automatically.

To help more organizations explore that possibility, OpenHands is making OpenHands Teams available as a self-service offering, enabling developers and engineering teams to get started with the platform without the complexity of a traditional enterprise procurement process.

Teams can begin with a focused application, such as vulnerability remediation, and gradually expand into more sophisticated automated engineering workflows.

For OpenHands, the larger ambition is to make software factories a practical part of how organizations build, maintain, and secure their applications.

And in a world where software development and cyber threats are accelerating, the ability to fix problems as quickly as they are discovered could become increasingly valuable.

Tags: Agentic AIAI Coding AgentsArtificial intelligenceAutomationCybersecurityOpenHandsSoftware EngineeringVulnerability Management
Editorial

Editorial

Next Post
Terra Security and Former Google CISO Gerhard Eschelbeck Examine How AI Could Reshape Cybersecurity

Terra Security and Former Google CISO Gerhard Eschelbeck Examine How AI Could Reshape Cybersecurity

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

About Us

Welcome to The San Francisco Tribune, your premier destination for business, technology, and culture. Our team delivers rigorously researched reporting, thoughtful analysis, and insightful commentary on the topics shaping industries, markets, and society.

  • Home
  • About Us
  • Contact us
  • DMCA
  • Privacy Policy
  • Subscribe to our Newsletter

© 2026 The San Francisco Tribune. All rights reserved.

No Result
View All Result
  • Home
  • Art
  • Business
  • Entertainment
  • Sports
  • Food
  • Magazine
  • Podcasts
  • Politics
  • Tech
  • Wellness

© 2026 The San Francisco Tribune. All rights reserved.