Qovery, a DevOps automation platform that makes software deployment effortless across any cloud environment, today announced the release of its AI DevOps Copilot, an AI agent that delivers answers, executes complex operations, and anticipates what’s next.
To date, it’s the only automation platform to cover the entire DevOps lifecycle.
I spoke to Romaric Philogène, CEO and cofounder of Qovery, to learn all about it.
From automation to intelligence
Simply put, DevOps is a collaborative approach that combines cultural principles, practices, and tools to automate and streamline the processes between software development and IT operations teams.
Qovery’s DevOps automation platform lets developers deploy and scale applications on their cloud — simplifying and automating infrastructure so tech teams can focus on what matters most: building great products.
According to Philogène, while Qovery already automates key steps such as provisioning, application deployment, and server security, no matter how much you automate, there are always edge cases — configuration issues that arise from how developers set up applications or infrastructure.
"That’s where AI becomes powerful,” he contends.
“Our Copilot lets software engineers continue working in the language they’re used to, without switching between interfaces. Instead of troubleshooting manually, they can simply say, ‘Copilot, can you fix this issue?’
The Copilot diagnoses problems, accesses configuration data, and resolves issues autonomously.
A new DevOps interface: natural conversation to reduce worker inefficiencies
Atlassian’s 2025 DevEx reports 50 per cent of developers lose 10 or more hours a week on non-coding toil in the software pipeline.
Qovery’s Copilot aims to change this: engineers give instructions in simple language and the system executes, turning a drain on resources into a competitive advantage.
According to Philogène, the Copilot represents a new type of interface—natural language. Instead of working through UIs, CLIs, or APIs, users can simply talk to the system.
“Our goal is to optimise how customers already use Qovery: reducing the time spent debugging, troubleshooting, or optimising configurations.”
For example, a team can request simple instructions that unused environments be shut down at the end of the day, or that a new service be deployed at a specific time, or more complex ones - such as “to run dev environments only from Monday 8 a.m. PT to Friday 6 p.m. PT; tear down non-used environments automatically at night; and trigger daily integration tests only if the platform has been unstable in the last 48 hours” and Qovery DevOps Copilot will handle the process automatically.
Developers are increasingly using AI not just to write code, but to verify it—to double-check what the model produces. This inspired Qovery to build a compatibility matrix: a list of tasks that the Copilot is guaranteed to perform safely.
“Of course, you can ask it anything, but we ensure that what’s officially supported works reliably without excessive human supervision,” explained Philogène.
Five AI agents powering the future of DevOps
What makes the company distinct is learnings from Qovery’s five years of experience streamlining the entire DevOps lifecycle — thousands of clients over 25 million applications and 30+ million infrastructure operations —- are built into DevOps Copilot’s five AI agents:
- AI Provision Agent: Automates environment creation and scaling by interpreting natural-language requests to allocate resources on demand, enforce quotas, and optimise infrastructure usage.
- AI FinOps Agent: Improves efficiency by detecting idle environments, scheduling shutdowns, and identifying opportunities to reduce spend.
- AI Observability Agent: Enhances incident response by retrieving logs, identifying anomalies, and suggesting solutions before issues escalate.
- AI DevSecOps Agent: Strengthens governance by enforcing compliance, requiring confirmations for sensitive actions, and maintaining a full audit trail.
- AI CI/CD Agent: Simplifies deployment and pipeline management with natural-language prompts for scheduling releases and managing test queues.
Strong guardrails for data security
While prompts are processed through Claude AI, (Anthropic's model) but private data and credentials are never transmitted. Qovery already runs customers’ infrastructure, so the company is very mindful of privacy. Philogène explained:
“We’ve designed guardrails so users can configure how the Copilot interacts with their environment.
For example, they can restrict it to read-only mode—where it reviews configurations and provides recommendations—or grant write access for direct operations.
Even then, the Copilot always requests confirmation before executing any changes. We ensure that only non-sensitive information is ingested into the AI system, and we’re continuously refining how we handle and anonymise data. Data safety is absolutely central to this product.”
Further, all actions are bound by role-based permissions, requiring explicit approval internally for high-stakes decisions like deleting databases or applications.
In the future, customers will be able to bring their own models for full control and compliance.
But for now, because Qovery controls both input and output data flows, the Copilot can learn from users' behaviour across organisations. It adapts to their workflows and infrastructure patterns, allowing for increasingly personalised recommendations.
“We have a huge advantage: access to well-structured data and consistent user behaviour patterns. This gives the AI the right context to make meaningful decisions without compromising privacy,” shared Philogène.
While too many tools in the software pipeline impact developer efficiency, engineers don’t want the fewest tools — they want the best tools for the job. Copilot, therefore, is totally agnostic — working seamlessly with any tool currently employed by development teams.
Further, in production DevOps, reliability and latency are everything. Philogène explained that the company’s new AI Copilot maintains real-time responsiveness by operating directly within the Qovery platform rather than relying on third-party intermediaries.
“That’s the beauty of the Qovery AI Copilot. It leverages the full capabilities of our platform, integrating directly with our APIs,” he said.
From automation to agentic autonomy
In terms of what’s next in the DevOps space, Philogène contends that the industry is moving toward fully autonomous agents.
“The Copilot will eventually identify and resolve problems before the engineer even needs to intervene. We’re not there yet—we still need to solve problems like hallucination, data ingestion limits, and context window size.
At Qovery, we mitigate these by feeding our models highly structured operational data and monitoring user behaviour to limit hallucinations.
Once those foundational challenges are solved, we’ll see agents capable of managing complex systems end-to-end.”
However, he doesn’t believe AI will replace developers. Rather, it will redefine their role.
“Developers are becoming more like product engineers—people who define and solve problems, not just write code.
Similarly, designers and engineers will use AI to bring their product ideas to life faster. It’s a complete shift in how teams collaborate.”
Two early pilot customers already tested Copilot during its pre-launch phase. Their feedback informed Qovery’s development roadmap.
Would you like to write the first comment?
Login to post comments