SiteGround Builds an AI Layer on Its Google Cloud Hosting

For most of its two decades, web hosting was judged on a short list of numbers: uptime, page speed, storage. SiteGround, the hosting company that says it serves more than 3 million domains, is betting that the next measure is how much work the platform does for its customers. Over the past year it has layered a stack of AI tools on top of its Google Cloud infrastructure, from a chat-driven WordPress agent to an app builder that turns a plain-language description, or an existing design file, into a working web application.

A cloud foundation built for continuous change

SiteGround began moving its hosting onto Google Cloud Platform in 2020, and today it runs, by its own account, 11 data centers across North America, Europe, Asia and Australia, backed by an in-house content delivery network with more than 170 edge locations, according to the company’s data center page. Customers pick a location at sign-up, and every site, store and web app in that account runs there.

The move was about more than renting servers. A Google Cloud case study on SiteGround, published after the company won a 2024 DORA Award, describes a gradual refactoring away from a monolithic architecture toward cloud-native services such as Cloud SQL and Memorystore. Google reports that SiteGround went from 12 deployments per week to 6 per day, cut its change failure rate from 5% to 0.75%, brought recovery time for a failed deployment down to 8 minutes, and shortened lead time for changes from 20 days to 8 (Google Cloud). That engineering cadence is the quiet enabler behind the rapid run of product launches that followed.

In a May 2025 post, CEO Nikolay Todorov framed the platform as ready for the AI era, pointing to Google’s network of more than 202 points of presence and claiming sites can load up to five times faster once Google Cloud is combined with SiteGround’s own caching and optimization layers (SiteGround blog). The company also notes that its data centers run on Google Cloud infrastructure that matches its electricity use with renewable energy.

AI Studio: many models, one workspace

The first major AI step came in December 2025 with SiteGround AI Studio, a workspace where customers can chat with leading models including ChatGPT, Claude and Gemini, generate images with Google’s Nano Banana model, and call on more than 15 specialized agents for jobs such as ad copy, social posts, SEO and translation (SiteGround). Every SiteGround customer receives a free Essential plan with 20,000 tokens a month, while a Plus plan raises the allowance to 100,000 tokens and unlocks the full roster of agents and a prompt library.

The design choice is notable. Rather than building one in-house model, SiteGround acts as an aggregator, letting small businesses compare how different models handle the same task without juggling separate subscriptions. The same token pool also powers AI writing inside its website builder, email marketing and ecommerce products.

From suggestions to actions: the WordPress and store agents

In February 2026 SiteGround launched its AI Agent for WordPress, which connects directly to a WordPress installation and carries out changes rather than just generating text (SiteGround). Through a chat window, site owners can create and publish posts, restructure categories, fill in missing SEO metadata across hundreds of images, update plugins, run bulk WooCommerce price changes, or pull sales reports. Agencies can apply the same instruction across many sites at once, and the agent also works with WordPress sites hosted elsewhere.

High-impact actions such as deletions, bulk edits and core settings changes sit behind a “Power Mode” that is off by default, so the owner has to opt in before the agent can make sweeping changes. The idea extended to commerce in mid-2026 with the AI Store Agent inside SiteGround Ecommerce. According to the September 2026 announcement, merchants can ask about figures such as average order value or top-selling categories and have the agent create products, brands and coupons on request (Digital Journal / Newsfile). “We built the AI Store Agent to make those day-to-day interactions with store data and operations more accessible,” said COO Reneta Tsankova.

Coderick AI: vibe coding with hosting attached

The most ambitious launch is Coderick AI, an app builder announced in March 2026 that joins the “vibe coding” trend of building software through conversation. SiteGround’s pitch is that most such tools stop at a prototype, while Coderick produces apps that are production-ready and hosted on its own infrastructure from day one, covering uses from business websites to internal CRMs and dashboards (PR Newswire). “We are taking the power of custom software development, once reserved for those with deep pockets and technical teams, and putting it directly into the hands of small business owners,” Todorov said.

The newest addition, announced November 9, 2026, is Import Design. Users can upload ZIP or HTML exports from Claude Design or Google Stitch, describe features such as user accounts, forms or bookings, and have Coderick generate the front end, back end and database (Newsfile). Projects can then be refined through chat, a visual editor, or direct edits in a code tab, which SiteGround opened up in August 2026.

The website builder has followed the same path. A major new version in March 2026 added a grid-based drag-and-drop editor, built-in ecommerce and product feeds to channels including Google Merchant Center and ChatGPT (SiteGround). In October 2026 an AI Wizard arrived that turns a typed or spoken description into a full site or store, showing an editable summary of name, goal, tone and color palette before it builds (SiteGround).

What it means

SiteGround’s trajectory illustrates a broader shift in managed hosting: the infrastructure is becoming a commodity, while the value moves up the stack to tools that remove labor. By running on a hyperscale cloud, the company can spend its engineering effort on agents and builders instead of hardware. For small businesses, agencies and internal teams, the practical result is that routine site administration, store updates and even simple custom applications can increasingly be handled by describing the outcome rather than doing the steps.

There are open questions. Token-based pricing makes heavy use harder to predict, and SiteGround itself recommends taking a manual backup before major agent work, since the WordPress agent does not create one automatically. Organizations weighing these tools should pilot them on non-critical sites, keep human review in the loop, and watch how output quality holds up at scale.

Conclusion

SiteGround has turned a long-running Google Cloud migration into a launchpad for an AI-first product line, moving in under a year from a multi-model chat workspace to agents that act on live sites and a builder that converts designs into working apps. Whether customers embrace AI that does the work, not just suggests it, will shape how hosting companies compete next.

Sources