How Smartsheet Uses Claude to Ship 3x More Code and Cut IT Costs by 60%

Smartsheet’s platform is embedded in the daily operations of more than 120,000 organizations, including over 85% of the Fortune 500. When the company decided to bring conversational AI to its customers and its own workforce simultaneously, the stakes were high: any AI deployment had to meet the same governance, auditability, and data residency standards that enterprise clients already relied on. The challenge was not just technical capability but institutional trust.

Internally, the engineering organization faced a version of the same problem its product was built to solve. AI coding tools available at the time stopped at line-level code completion. Data lived in silos across Snowflake, service databases, and internal APIs, and no single tool could reason across those boundaries. SVP of AI & Data Platform Drew Garner framed the gap directly: “Here’s the thing most companies get wrong about AI adoption: they think it’s a tools problem. It’s not. It’s a knowledge problem.”

Smartsheet evaluated multiple foundation models for each use case and selected Claude Sonnet as the core model for customer-facing surfaces and Claude Code for engineering. Routing all API traffic through Amazon Bedrock gave the security and compliance team full visibility and control without rebuilding governance infrastructure. The Smartsheet MCP Connector for Claude launched in March 2026, supporting 36-plus operations and allowing customers to view, create, and manage their Smartsheet work directly through conversation. The engineering team built a nine-agent SDLC template covering architecture planning, implementation, code review, merge request processing, and accessibility remediation, and deployed Claude Enterprise to every employee on the same day as the company’s fiscal year kickoff.

The scale of adoption was faster than anyone anticipated. The MCP connector reached 1,400 organizations within 30 days, generating 1.76 million tool calls. Critically, 48% of those interactions went beyond querying data: users were creating sheets, updating rows, and flagging project risks through conversation rather than navigating the product UI. Inside engineering, documentation that previously took two weeks now takes four hours. Cycle time across the organization dropped 28%, and the highest-performing engineers are shipping five to seven times their previous output. Claude-powered agents now handle 30% of IT helpdesk tickets automatically, and prompt caching on the API reduced both cost and response latency.

The broader lesson Smartsheet is drawing from the rollout is about organizational knowledge, not tool selection. Because Smartsheet already holds customers’ business logic, project structures, and workflow data, connecting Claude to that institutional context produces an AI layer that already understands the business. The company is now expanding the MCP server toward full product parity and growing its internal connector ecosystem from five integrations to more than twenty.