Agentic Cloud: Reinventing the Cloud with AI Agents

AI is shaping up to be one of the most transformational technologies of our lifetimes and will change how we live and work just like the PC, the internet, and the smartphone did. The launch of ChatGPT was a seminal moment in the evolution of AI, and since then, AI’s adoption has accelerated. According to “The rapid adoption of Generative AI. Bick, Blandin, & Deming (2024)”, AI is being adopted at twice the rate of the PC and internet at a similar point in their evolution.

Not just adoption, the rate of innovation in AI has been faster than anything we have seen before. So where is AI headed?

To answer this question, let us start by looking at how other transformational technologies have evolved throughout history. Take the Internet for example – it started with the infrastructure - with routers and network switches. But the true inflection point came with the emergence of platform capabilities, such as web browsers and HTML, which allowed people to build and use applications like search engines, e-commerce, and social media.

If you look at the PC, microprocessors and memory and other infrastructure components came first. But what triggered the massive adoption was the creation of a platform – WinTel, which provided a standardized environment for developing and running PC applications. Same with the smartphones; the infrastructure started first with modems and cell towers, but it really took off when a platform enabled the creation and use of applications.

AI will follow a similar IPA (Infrastructure -> Platform -> Application) model, with the innovation moving from Infrastructure (e.g. GPU) to Platforms (e.g. DigitalOcean GenAI Platform) and Applications (e.g. Copilots), and AI agents are going to accelerate this transformation.

What are AI agents?

In essence, an AI agent is a piece of software that can act like humans – take input, reason about things, and perform actions. Agents are really four sequences of primitives. The agent gets a command from the user to perform some action, the agent then gets the information it needs by acquiring data from the outside world, it then reasons and plans how to get the job done, and finally it performs the actions.

AI agents will transform businesses

Agents will drive a massive automation of work.Consider the example of a customer service agent today. Many of the calls that such an agent handles can be handled by an AI agent. An average call interaction handled by a human agent that takes about 10 minutes today costs around $3^, while an AI agent can do that for one cent^^. Not just that, the quality of these AI models is improving exponentially and the cost of AI is decreasing by 10X YoY. So as you factor in the progress over the next few years, the conclusion is inevitable - agents will be automating or assisting with a lot of the “human work”. That means, businesses of all kinds will have no option but to reinvent themselves with AI agents in order to serve their customers in the best and most cost effective way possible.

According to Statista, annual US salaries total approximately $200 billion for HR, $330 billion for IT, and $1.1 trillion for sales & marketing, and as a lot of enterprise workflows get automated, the cloud and AI TAM will increase by hundreds of billions of dollars as businesses deploy agents to automate enterprise work.

At DigitalOcean, we will fully participate in this TAM expansion by enabling customers to build and deploy agents through our GenAI platform.

Agentic Cloud - AI agents transforming the cloud

Given that agents will touch every enterprise workflow, it is fair to ask “how will agents change the cloud”? Indeed, we are at an inflection point for the cloud due to two emerging paradigm shifts: 1) Natural language as code, and 2) agents automating human work. The first will change how developers interact with the cloud, while the second will automate many DevOps/CloudOps and other cloud workflows. Like in previous transformations, there will be winners and losers. The winners will be the ones who use the new paradigms to reinvent themselves, while the losers will be the ones who do not change.

At DigitalOcean, we will be at the forefront of reinventing the cloud by embedding agentic automation into our services. Let me illustrate how we are applying the Agentic cloud paradigm to reimagine web hosting. Cloudways is a web hosting software from DigitalOcean that is available on multiple clouds - DigitalOcean, AWS, GCP, Vultr, and Linode, Cloudways was recently ranked as the No. 1 web hosting software by CNET, and is one of the most popular web hosting applications for WordPress, which powers more than 40% of websites.

Customers tell us that dealing with system issues (disk usage, DDOS, etc.) is a significant pain point in web hosting and they often spend several hours a month debugging and fixing such issues. Today, DigitalOcean receives more than 600,000 customer support requests a year for hosting system issues, with each of these support requests spanning 40 minutes on average. Therefore, we asked the question - “How would Cloudways have dealt with system issues if AI and agents were available when it was first built”? The answer was pretty obvious – Cloudways would use agents to autonomously rootcause and fix these issues, and enable customers to interact in natural language to gain insight about these issues. So, we set out to do exactly that by embedding SRE (system reliability) agents inside Cloudways that can take input from users in English and autonomously debug and fix system issues.

We launched Cloudways AI agent – one of the most sophisticated uses of GenAI in the industry - in private preview with about 250 customers in Q1 ‘25. As shown in the video below, Cloudways AI agent provides troubleshooting insights and 1-click automated issue resolution. The user only needs to click the alert when issues arise and the AI agent instantly explains the issues and provides remediation steps. And with a single click, AI jumps into action and fixes the issues automatically. Users no longer need to muck around with code, and all the interactions are done in natural language. What used to take an hour now just takes a minute, saving businesses thousands of dollars a year^^^.

We will be releasing the General Availability (GA) very soon and are encouraged by the incredible feedback from early customers - “Cloudways Copilot has been a game-changer”; “It is the first good implementation of AI I’ve seen”; “It has saved us countless hours”; “It is incredibly fast, accurate, and intuitive; “ The insights it provides are spot on and the recommended steps have been 100% correct every time.”

This is only the beginning. We are going to embed AI agents in all of our services, providing customers a truly differentiated cloud experience. As an agentic cloud, DigitalOcean - the simplest, scalable and approachable cloud – gets even simpler, more scalable and approachable. We believe we are uniquely positioned to lead the next wave in reimagining the cloud and I will discuss more in a future blog.

^ Based on U.S. average customer service salary of $20/hr for a 10-minute call.

^^ Based on agent pricing of $1 per million tokens.

^^^ Based on average customer spend of $200 per month on these tasks.