Arnas Donauskas, product manager at Hostinger, described on the Jukebox Podcast how AI is being used to simplify WordPress site creation, monitoring, troubleshooting, and performance tuning. His path from building a college website to leading tools for onboarding and site maintenance informs the product choices Hostinger is making.
Onboarding and proactive operations
Hostinger applies AI in two main ways: as an onboarding assistant that can create sites from user prompts via a WordPress AI builder, and as a proactive operations assistant that watches hosted sites continuously. The platform performs ongoing HTTP checks and, when it detects an error (for example 403s, 500s, or other critical failures), it collects logs, removes sensitive information, and feeds the sanitized data into a troubleshooter that maps likely fixes to available automated actions.
Automation and human fallback
Where possible, Hostinger applies fixes automatically rather than only suggesting them. Current results show roughly a 70% success rate for first-attempt automated fixes, translating to about 16,000 sites repaired per month. Additional automated attempts raise the overall success rate further. When automation cannot resolve an issue, Hostinger’s support specialists receive a comprehensive dossier of logs and attempted fixes so they can intervene more quickly. Backups and safe restore options are available to protect sites in case a rollback is needed.
Performance optimization
Performance work is driven by benchmarks (Google PageSpeed) run per site. Hostinger applies staged, safe optimizations tailored to each site: smarter image handling, JS/CSS minification, caching tweaks, and other conservative changes first. Early results show average improvements around +20 PageSpeed points on mobile and +10 on desktop. Customers are then offered optional, higher-risk adjustments (for example some aggressive lazy-loading changes) with clear rollback paths if visuals break.
Personalization and future granularity
Optimizations are individualized per site today, but Hostinger is working toward more granular decision logic that understands site types (brochure, blog, WooCommerce) and page roles (landing page vs. product page). There isn’t extensive user-facing customization for each individual optimization yet, but the roadmap includes enriching the AI’s knowledge of different site architectures so recommendations and automated actions can be more targeted.
Communication and user control
After changes, clients receive an impact summary and a plain-language log of actions. Hostinger favors in-dashboard review over frequent email alerts; many users review the results in 10–15 minutes. Advanced users may be given downloadable, detailed logs later. The product emphasizes keeping customers in control: fully automated modes could exist, but confirmations and clear, trustworthy choices are important because these are business-critical sites.
Engineering, models, and privacy
Hostinger tests multiple models and providers to balance latency, cost, and accuracy; the team measures both speed and correctness when selecting model providers. While many tasks can run on free or open-source models with the right prompts, paid proprietary models often offer better accuracy or performance. Development of the troubleshooter and optimiser required nontrivial engineering effort—about seven to eight months—and these capabilities are included in hosting plans at no extra charge.
Roadmap highlights
– Raise automated first-attempt fix success toward a 90% target by expanding supported automatic fixes.
– Build a “way back machine” to track file changes, identify the exact change that caused a failure, and offer safe restores or targeted rollbacks (including estimating missed e-commerce orders).
– Extend the troubleshooter and optimiser to softer issues: SEO checks, accessibility fixes, image restoration, grammar corrections, and in-admin suggestions.
– Offer risk-tiered optimizations where customers can approve more aggressive changes knowing they can revert.
Monitoring approach and practical notes
Hostinger uses HTTP monitoring as the primary signal for critical issue detection because customers run on Hostinger infrastructure; this gives fast, reliable detection in most cases (with some caveats around certain CDN setups). For performance, scheduled PageSpeed tests trigger optimization suggestions when meaningful gains are available.
How to give feedback
Arnas encourages users to send feedback through Hostinger’s support chat and connects on LinkedIn. The overall goal is to automate routine maintenance and safe fixes while keeping owners informed and in control, reducing the time and stress of site upkeep so businesses can focus on growth rather than maintenance.