In a nutshell: Smartsheet operates an MCP server on AWS that provides AI agents with structured access to platform data and has saved 3 billion tokens to date through token optimizations.
Smartsheet has built a remote Model Context Protocol (MCP) server on AWS to give AI agents direct access to the platform’s data and functions. The system enables language models and autonomous agents to work with Smartsheet data while shortening workflows by weeks.
Smartsheet, an enterprise platform for project management and workflow management, has developed an MCP server architecture that empowers AI assistants such as Amazon Q and Claude Desktop, as well as its internal system Smart Assist, to access Smartsheet functions. This enables AI assistants to analyze project data, update tasks, create sheets, and manage workspaces — all through natural language interfaces. The system also supports autonomous AI agents that operate without human input and can coordinate through the MCP interface — for example when capturing requirements, selecting tasks, or attaching test results.
The architecture leverages multiple AWS services in the critical data layer: AWS Fargate with Amazon ECS for stateless containers, Amazon Kinesis Data Streams and Apache Flink for event-driven data ingestion, and Amazon Bedrock and Amazon Neptune for language model inference and knowledge graph-based cross-project analytics. Requests pass through an API Gateway layer with AWS WAF, Shield, and Application Load Balancer before reaching the MCP server. From there, transactions are executed via Smartsheet APIs and insights are retrieved from an intelligence layer built on Databricks and Neptune. Change events flow through Kinesis and Flink into an S3-backed intelligence layer following the medallion architecture pattern.
A central optimization feature is the MCP server’s AI-optimized interface, which is designed to minimize token costs and reduce hallucinations. According to the company, Smartsheet has saved over 3 billion tokens through these optimizations since launch. A single MCP layer serves both internal and external agents with the same infrastructure, the same tools, and the same optimizations — a deliberate architectural design that means development happens once and benefits all clients.
AI traffic patterns differ fundamentally from classical request patterns: agents orchestrate sequences of tool calls in seconds, while the language model then pauses and reasons through solutions. Smartsheet uses ECS Auto Scaling with target tracking policies to handle both sudden spikes and sustained throughput. This scaling strategy is specifically tailored to the bursty nature of agent traffic.
Source: aws.amazon.com · Published 17 July 2026
Lumi AI News — AI-assisted curation in accordance with Article 50 EU AI Act. Paraphrase and classification via Lumi News Pipeline v1.7.3.