Thanks to everyone who joined today’s Airtable AI Builder Crew: Actionable Insights with AI!
Today we went beyond “where can I add an AI field?” and looked at how to design Airtable systems where AI can actually help people make better decisions.
The big idea: AI features ≠ an AI system.
AI becomes much more useful when it has structured context to work from. Using a project management example, we walked through how to:
🔹 Structure projects, tasks, RAID items, people, and time at the right levels of the data model
🔹 Use formulas and rollups to create AI-ready context from information distributed across related tables
🔹 Turn that context into focused AI agents for project health, risk categories, delivery confidence, client engagement, and more
🔹 Use constrained outputs like single selects and multi-selects so AI-generated insights can actually power charts and dashboards
🔹 Layer agents so one analysis can become context for the next
🔹 Decide when an Airtable field agent makes sense versus when an MCP-connected agent is useful for pulling context from outside Airtable
🔹 Build interfaces that let users move from portfolio-level signals into the underlying evidence
One of my favorite takeaways from the demo: dashboards are much more powerful when they’re a byproduct of the work instead of another reporting exercise.
If the actual work, decisions, risks, approvals, KPIs, vendor activity, contracts, or other operational processes are happening in Airtable, you’re continuously creating the structured evidence AI needs to identify patterns and surface insights.
And while we used project management for the demo, the same pattern applies to vendor management, contract management, KPI management, intake and approvals, and plenty of other operational workflows.
Thanks again to everyone who joined, asked questions, and experimented with us!
What would you like us to dig into at the next AI Builder Crew?