I’m trying to understand the practical decision point between Smartsheet’s native AI capabilities, Copilot, MCP and custom API development.
For a project management use case, I could imagine the same requirement being approached several ways:
Native Smartsheet:
Build reports, dashboards, formulas and automations directly into the solution.
Copilot:
Use natural language to interpret or work with information already in Smartsheet.
MCP:
Give an AI agent structured access to Smartsheet so it can reason across project information and potentially interact with other tools.
Custom API:
Build deterministic integrations where we need precise control over data, logic and system behaviour.
What I’m still trying to establish is the decision framework.
For example:
- When is MCP genuinely better than a report or dashboard?
- When should a task remain deterministic rather than agent-driven?
- What types of workflows are good candidates for MCP?
- Where does Copilot end and MCP begin?
- What are people doing when they need MCP to work across Smartsheet plus other enterprise systems?
I’d love to hear from people who have actually implemented these patterns rather than just experimented with them.
