Self-healing infrastructure is the promise every cloud hosting platform is making in 2026. Cloudways, a managed cloud hosting platform now part of DigitalOcean, is making it too. With one exception.
Copilot, Cloudways’ AI infrastructure diagnostics tool, can identify a DDoS pattern, a disk space constraint, or a database query problem and surface the root cause in minutes. It can recommend a fix. What it will not do is apply that fix without the customer clicking to approve it. That is not a technical limitation. It is how the product was designed, and Ayaz Ahmed Khan, Senior Director of Engineering at DigitalOcean, is specific about why.
The first time we spoke with him, the focus was on how Cloudways built Copilot: the data it relied on, the engineering decisions, and what the product actually does at the infrastructure layer. The questions that only have answers once real customers start using it – adoption curves, pricing decisions, what the AI handles confidently and where it still hands off to humans – those we saved for this conversation.
Part Two: Adoption and Business Impact
webhosting.today: What has the adoption curve looked like across the 100,000+ servers on the Cloudways platform? What percentage of Cloudways customers actively use Copilot today?
Ayaz Ahmed Khan: We’ve been very encouraged by Copilot’s adoption since launch. Strong early interest has evolved into broader, trust-led adoption as customers begin using it in real-world environments and share their experiences.
We plan to expand the rollout to all customers in the near future. During the preview phase alone, we onboarded nearly 400 customers, giving us a diverse cross-section of real-world environments to validate Copilot’s performance at scale.
We’ve also seen usage evolve. Customers often start by using Copilot for visibility into what’s happening across their servers, before gradually relying on it more proactively to help diagnose and resolve issues.
webhosting.today: Copilot’s SmartFix feature offers one-click remediation for issues like bot crawling, disk space restrictions, DDoS/DoS attacks, and database query problems. What is the ratio of issues Copilot resolves with SmartFix versus issues it diagnoses but escalates to human support? Are there categories of problems where Copilot consistently fails?
Ayaz Ahmed Khan: Diagnosis is one of Copilot’s key strengths, particularly when it comes to high-volume, pattern-based problems – such as resource constraints, bot traffic, or database inefficiencies – that show up consistently across environments.
In practice, customers can move from identifying an issue to understanding the root cause in minutes, rather than spending hours investigating or escalating to support.
What we’ve seen so far is that Copilot reliably identifies issues, but customers don’t always choose to apply the fix immediately. This is either because they want to validate it themselves or because they’re still building confidence in automated remediation.
There are also categories of problems that remain more challenging to diagnose and remediate. Highly customised application-level issues, for instance, where the signals are less consistent, are one area where human expertise still plays a vital role.
Currently, we are focused on server-level issues, which are already in production, with a near-future release set to expand investigations into application-level issues.
webhosting.today: SmartFix currently requires the user to click to apply a fix. Is the long-term direction for Copilot to act autonomously – detecting and resolving issues without human approval – or will it always recommend and wait? For production environments, an automated fix can cause more damage than the original issue – how do you balance speed of resolution against the risk of autonomous action?
Ayaz Ahmed Khan: This is something we’ve been very deliberate about from the outset. While the long-term vision for AI in hosting is clearly moving towards more proactive and self-healing infrastructure, we’ve taken a human-in-the-loop approach with Copilot by design.
In practice, this means SmartFix requires explicit user action. We know that production environments carry real risk, and even the best-intentioned automated fixes can have unintended consequences if applied without context. What SmartFix does well is separate diagnosis from execution. It uses AI to analyse signals, identify the root cause, and recommend a fix, but the actual remediation is carried out in a controlled way. In our experience, customers are far more likely to adopt automation when they understand what’s happening and feel in control of the outcome.
We see a path towards more automated workflows, particularly for well-understood, repeatable issues, but this will be introduced in a measured way. The systems will have to operate within clearly defined boundaries rather than acting autonomously without oversight. We don’t want to remove the human from the loop entirely, but rather give them the ability to move faster and make better decisions. That, in our view, is the real value of AI: taking on the heavy lifting behind the scenes.
Eventually, in an ideal state, customers should not have to worry about issues affecting their server or site. The system should be intelligent enough to identify problems, find the root cause, and reliably and safely fix them. That’s our long-term ambition.
We do have customers who want us to automatically fix common issues, while others specifically do not want the system to make changes automatically. Over time, we believe more customers will come to trust the system’s ability to identify and resolve issues, and gradually hand off more of that responsibility.
webhosting.today: Hostinger’s AI assistant Kodee now handles 81% of customer support interactions and saved the company approximately EUR 9 million in 2025. Has Copilot had a measurable impact on Cloudways’ support costs or team size? Can you share specific numbers?
Ayaz Ahmed Khan: Since launch, we’ve continued to see gradual adoption of Copilot among users who have it enabled, alongside growing use of the one-click SmartFix functionality to resolve issues.
One of the most immediate impacts has been on diagnosis time. Issues that previously took 30-60 minutes to investigate can now be analysed by Copilot in a matter of minutes, surfacing both the root cause and a recommended fix. This fundamentally changes how and when support teams need to engage.
webhosting.today: Copilot uses a credit-based pricing model, with paid tiers at $9.99 and $19.99 per month. What drove the decision to make AI diagnostics a paid add-on rather than including it in the base Cloudways hosting price? Which tier sees the most adoption, and does the credit model affect how frequently customers use the tool?
Ayaz Ahmed Khan: From the outset, we wanted to clearly demonstrate the value Copilot delivers: reducing troubleshooting time, improving visibility, and helping customers resolve issues faster.
The decision to introduce it as a paid add-on reflects that value proposition. Rather than bundling it into the core platform, we wanted to give customers the flexibility to adopt it based on the value they expect to get from it, particularly as usage can vary significantly depending on the size and complexity of their environments.
As mentioned, customers tend to engage with Copilot during periods of higher activity, rather than using it as a constant, always-on tool. With that in mind, we aim to strike the right balance between cost and quality, and continuously look for ways to optimise pricing without compromising the quality of the insights.
webhosting.today: Cloudways has historically been strong with agencies – team management and client-level billing. How does Copilot change the way agencies manage infrastructure for their clients? Can an agency use Copilot across all client servers from a single view, or is it per-application? Is there an agency-specific pricing model?
Ayaz Ahmed Khan: For agencies, a big part of Cloudways’ appeal is that infrastructure is managed at a day-to-day level. This means that, instead of relying on reactive support or manually digging through logs, teams can quickly understand what’s happening across their environments and resolve issues much faster.
When you’re managing multiple client sites, that’s invaluable. Copilot helps to reduce the time spent troubleshooting individual issues, which can quickly add up across dozens or even hundreds of applications, and allows teams to stay focused on delivering value to their clients rather than maintaining infrastructure. This helps them to focus on their own areas of expertise.
In the near future, our roadmap is packed with features designed to enhance Copilot, including new ways to manage more of the platform through GenAI and natural language interactions. These offerings will directly benefit agencies by providing simplified workflows to manage servers and applications across their accounts from a single place, without having to worry about the intricacies of handling those tasks manually.
webhosting.today: Copilot focuses on infrastructure diagnostics and automated remediation – a different layer from the AI site building tools entering the WordPress hosting space, like Bluehost’s WonderSuite, WordPress.com’s AI Assistant, or white-label solutions like Extendify that hosting companies can integrate into their WordPress offering. Is infrastructure diagnostics the long-term focus for Cloudways AI, or are there plans to expand into the WordPress application layer – whether built in-house or through partnerships with companies like Extendify?
Ayaz Ahmed Khan: A lot of AI innovation in hosting today is centred around site creation, content generation, or design, which are important but relatively one-time interactions. Where we see a longer-term opportunity is in what happens after that – when sites are in production and performance, reliability, and uptime become critical.
Copilot’s focus on infrastructure diagnostics and remediation sits firmly within our area of expertise. Since the beginning, we’ve analysed real system signals to help customers identify root causes and resolve issues in real time. Copilot adds another layer of value by being much more tightly integrated into the day-to-day workflows of developers and agencies.
As Copilot grows and becomes more deeply embedded into the platform, there is a natural progression towards more application-level intelligence, particularly where it can provide meaningful, context-aware recommendations without adding complexity.
In addition, our roadmap extends into AI-powered site building and “vibe coding” solutions for customers who prefer to use AI to help set up their sites. However, our primary focus remains on the production side of the customer experience: improving the process of getting sites into production and ensuring they continue to run seamlessly as customers scale their businesses.
webhosting.today: DigitalOcean acquired Cloudways for $350 million in 2022, adding a managed hosting layer on top of its infrastructure products. With Copilot now adding AI diagnostics to that managed layer, how does this change things for Cloudways customers? Is the goal to justify higher pricing through AI operations, or to reduce churn by keeping customers on the platform longer?
Ayaz Ahmed Khan: For customers, Copilot means that, rather than relying on reactive support or manual investigation when something goes wrong, they have an intelligent system that can surface issues early, explain what’s happening, and guide resolution in real time.
This fundamentally changes the support experience. It reduces the operational burden on teams, making it easier to manage complex environments without requiring deep infrastructure expertise.
From a commercial perspective, it’s less about any single lever – whether that’s pricing or retention – and more about strengthening the overall value proposition. As customers see more day-to-day value from the platform, whether through time savings, improved reliability, or reduced reliance on support, that naturally supports both long-term retention and opportunities to expand usage.
Our ultimate goal is to provide the best possible hosting experience for our customers at highly competitive prices. AI naturally adds to these costs, as frontier-model tokens come at a premium. However, even as we integrate AI more deeply into Cloudways, our goal is to keep pricing as low as possible. We’re always focused on optimizing costs without compromising on quality.
