
Cloudways, a unit of DigitalOcean, has made available a public preview of a generative artificial intelligence (AI) capability for its managed hosting service.
Suhaib Zaheer, senior vice president for managed hosting at Cloudways, said the Cloudways Copilot leverages large language models (LLMs) to help troubleshoot issues. Later this year, the provider of managed hosting services will add additional AI copilot capabilities to automate application developer workflows.
Those capabilities should substantially level the IT playing field between smaller organizations that typically take advantage of managed Cloudways hosting services and larger organizations that rely on more expensive services provided by hyperscalers as the overall level of expertise required to deploy and manage applications in the cloud declines in the AI era, added Zaheer.
Itโs not clear to what degree AI will encourage smaller organizations to deploy more applications, but they will become easier to build and manage. As such, the expectation is that organizations will be able to deploy more software than ever without necessarily having to increase headcount.
AI, however, isnโt going to replace the need for IT administrators so much as it will provide them with a wingman capable of executing various tasks as directed, noted Zaheer.
The challenge, of course, is determining to what degree to rely on AI copilots to execute a task. The LLMs used to automate a task are probabilistic tools, which means they donโt always execute the same task the same way every time. Most IT workflows, in contrast, are deterministic in nature, which often require IT teams to document how they were specifically completed in line with a set of prescribed policies and procedures.
There is no doubt that AI copilots will reduce the overall level of toil experienced by IT teams. In fact, there may come a day soon when IT teams will not want to manage increasingly complex IT environments without relying to one degree or another on generative AI tools and platforms.
However, just about every task assigned to a generative AI tool will need to be reviewed by an IT administrator to ensure it was executed as required, especially given the tendency of an LLM to occasionally hallucinate in a way that convincingly surfaces erroneous summarizations and recommendations. If IT teams act on that advice, it wonโt be the AI copilot that will be held responsible for the outcome.
Ultimately, IT teams will be able to invoke a wide range of AI copilots and agents to automate tasks. Arguably, the next big challenge will be finding ways to orchestrate IT workflows being performed by multiple AI agents that are supervised by IT administrators. In fact, one of the benefits of AI may be a reduced need to rely as much on DevOps specialists to programmatically create those often-complex automation workflows.
In the meantime, IT teams should assess which tasks today are being performed by IT professionals that might just as easily be one day soon assigned to an AI agent, starting with the ones they never especially liked doing in the first place.