Malcolm Peralty, a WordPress veteran with nearly two decades of experience, joined the Jukebox Podcast to outline his journey from early WordPress blogging through project management and a period at Acquia/Drupal, then back to WordPress as a Technical Account Manager (TAM) at Pressable, a managed WordPress host running its own WP Cloud stack.
Background and role
Malcolm began using WordPress before version 1.0, worked as a full-time blogger, and later moved into agency and project management work. Time at Acquia broadened his enterprise hosting and Drupal experience, but better compensation and the right fit brought him to Pressable. As a TAM he focuses on site strategy and long-term planning—performance tuning, plugin and theme guidance, capacity planning, and helping teams choose the appropriate service tier—rather than tier-one support or direct sales.
Proactive support and resource stewardship
Pressable’s TAMs take a proactive approach: they regularly review site performance and resource usage and recommend changes that can reduce costs for customers and free capacity for the host. When deeper investigation is needed—issues that escalate beyond standard support or unusual resource consumption—the TAMs step in to diagnose and coordinate fixes. Because Pressable runs its own WP Cloud instead of relying directly on AWS or GCP, TAMs serve as a translation layer between WordPress-specific needs and the platform engineers responsible for the stack and hardware.
Performance trade-offs and cacheability
A central performance divider is cacheability. Static brochure sites typically are highly cacheable and run efficiently with minimal intervention. Sites with dynamic, user-specific interactions—e-commerce, LMS platforms, or carts—generate uncached traffic and require more planning and resources. Page builders and plugin bloat (Elementor, Divi, complex commerce integrations) are frequent causes of poor performance and cache-breakage. Pressable’s worker model assigns one worker per vCPU, giving each site a dedicated execution lane that favors predictable behavior under load compared with shared-worker approaches.
Emerging hosting trends
Malcolm highlighted several infrastructure priorities: WebAssembly experiments (for in-browser WordPress sandboxes and ephemeral staging) open new testing workflows; improved caching layers, database replication, and real-time streaming replication reduce data-loss windows and improve cross-region resilience; and better logging, auditing, and diagnostics will let admins track who changed what and when without depending solely on backups.
AI integration: Pressable MCP
Pressable is building MCP, an AI-driven control interface that exposes hosting control functions via APIs so natural-language commands can trigger real hosting tasks. Use cases include spinning up sandboxes from VS Code or GitHub with a plain-language prompt, finding and bulk-updating plugins across many sites, orchestrating code pushes and filesystem/database syncs, and then reporting results. The goal is to automate repetitive multi-site workflows, speed proofs-of-concept in client meetings, and reduce time-to-delivery for agencies and teams.
Guardrails, auditability, and human oversight
MCP is being introduced in phases with a human-first design. Planned guardrails include confirmation steps for destructive actions, robust backups (hourly DB snapshots and daily file backups), and protections like simlinked core files to avoid accidental deletions. Customers can choose AI backends (local or cloud models), with responsibility shared: Pressable will enforce platform-level protections while users should configure prompts and safety checks. Comprehensive audit logs and readable step-by-step records are essential so rapid AI-driven changes can be traced and selectively rolled back by humans.
Scaling, costs, and AI-driven traffic
Automated agents and AI crawlers can impose real costs by creating uncached sessions, performing repeated actions, or generating heavy traffic that evades basic blocking. Hosts face billing and operational challenges: should customers pay for indexing by external agents, or should providers absorb those costs and risk margin erosion? Potential mitigations include smarter blocking and rate-limiting, plus new pricing or usage models for AI-driven traffic and long-term audit storage.
Human relationships and the future of support
Malcolm stresses a human-first stance: AI should enhance workflows without replacing essential human judgment. Early MCP adopters will likely be agencies and dev teams comfortable reviewing complex changes. AI can accelerate routine diagnostics, remediation scripting, and generate assets like CSS, blocks, or plugin scaffolds, but generated artifacts must be audited for security, performance, and maintainability. Over time AI may handle more routine fixes while humans review higher-risk decisions.
Practical guidance
For small static brochure sites, Pressable’s platform will generally work without deep TAM involvement. For dynamic, uncached platforms (LMS, heavy commerce), plan for architecture consultations and possibly higher-tier resources. When delegating tasks to AI, maintain conservative prompts that require confirmation for destructive steps, keep thorough logs, and preserve backups so changes are auditable and reversible. Agencies and multi-site teams will see the fastest ROI from MCP features like bulk actions, quick sandbox creation, and automated deployment flows.
Closing
Malcolm views AI as an accelerant for hosting and workflows: when integrated thoughtfully, it will make hosts more valuable to agencies and teams that need speed and automation. At the same time, hosts must solve for auditability, cost impacts, and human trust to make AI-driven features sustainable.
More information
Pressable: https://pressable.com
Malcolm Peralty: https://peralty.com and LinkedIn/WordPress Slack communities for contact.