How Smartsheet built a remote MCP server on AWS

This brief explains how companies can build robust, scalable, and secure remote computing infrastructure using cloud services, even for specialized machine learning workloads. The article details Smartsheet's experience in developing a remote "Managed Compute Platform" (MCP) server leveraging AWS, focusing on the architectural decisions around security, governance, deployment strategies, and crucial AI-specific optimizations. It provides a blueprint for managing complex server environments at scale without needing to own and operate physical hardware. For an independent SaaS founder based in Austin, Texas, this approach means the ability to offer advanced features requiring significant computational power, like AI-driven analytics or real-time data processing, without the prohibitive upfront investment in data centers. They could, for instance, develop a new subscription tier for predictive inventory management, where small businesses in categories like retail or restaurants can upload transaction data for forecasting demand. This allows the founder to scale compute resources on demand, paying only for what's consumed, and focus on product innovation rather than infrastructure maintenance. Similarly, an internal IT team at a mid-size real estate firm in Chicago could adapt these principles to host a powerful internal tool for property valuation or market trend analysis. By using a similar remote MCP architecture, they could provide their agents with secure, on-demand access to computationally intensive models, improving decision-making speed without burdening the company's local network or requiring specialized in-house server expertise. To capitalize on this, consider replicating a small-scale remote MCP for a specific, computationally intensive task within your current workflow. Identify one process that currently bottlenecks due to a lack of local computing power or requires significant manual effort to manage, such as data sanitation, complex financial modeling, or high-resolution image rendering. Outline the essential security and governance requirements for this process, then explore a cloud provider's managed compute services to build a proof-of-concept environment this week.