AI Project Notification Triage Agent

How Cassidy automates project notifications using AI

Step 1: Trigger on new project notification

The Workflow activates when a new event arrives—RFI created, submittal status changed, change event opened, safety incident logged, or schedule variance detected—via API webhooks, email parsing, or chat connectors.

Step 2: Normalize and enrich the notification

Cassidy extracts key fields (event type, due date, action required by, attachments) and enriches them with contract context, CSI spec sections, WBS cost codes, schedule activity links, and the current reviewer matrix from your Knowledge Base.

Step 3: Classify and correlate events

Using AI classification, Cassidy identifies the artifact type and risk category—distinguishing blocking RFIs from informational ones, initial submittals from resubmittals, and high-severity safety issues from routine observations—while deduplicating alerts across systems.

Step 4: Calculate priority score and SLA timers

Cassidy scores each item against configurable factors: overdue status, critical path float, cost exposure thresholds, safety severity, and contractual review windows. SLA timers begin tracking, and escalation timeouts are set based on your project's rules.

Step 5: Route to the right stakeholder

Based on role-based routing policies, Cassidy delivers the notification to the appropriate party—PM for commercial risk, Superintendent for field blockers, Safety for incidents—via their preferred channel with deep links back to Procore or ACC.

Step 6: Escalate and nudge as needed

If items approach or breach SLA deadlines, Cassidy automatically sends reminders and escalates to project leadership. Human-in-the-Loop keeps your team in control of exceptions and ensures nothing critical slips through.