AI Engagement Document Extraction Agent

How Cassidy automates engagement document intake using AI

Step 1: Trigger on document arrival

The Workflow activates when engagement documents land in designated channels—email dropboxes, SharePoint folders, client portals, or e-sign completions—capturing MSAs, SOWs, NDAs, proposals, and change orders as they arrive.

Step 2: Classify and assemble document sets

Cassidy identifies each document type and builds the contract family structure, establishing relationships between parent agreements and child exhibits while detecting duplicates and version conflicts.

Step 3: Extract commercial and legal data

The Agent pulls key fields from the Knowledge Base-informed extraction model: party names, effective dates, fee structures, rate cards, liability caps, IP terms, milestones, and engagement-specific metadata like practice line and partner assignments.

Step 4: Normalize and validate against master data

Cassidy matches extracted counterparties to CRM accounts, standardizes currencies and date formats, and runs business rule checks—flagging deviations from playbook terms or policy thresholds that require attention.

Step 5: Human-in-the-Loop review for exceptions

Low-confidence extractions and policy exceptions route to a review queue where engagement managers can verify fields side-by-side with source snippets, approve corrections, and clear items for downstream processing.

Step 6: Route to CRM and CLM systems

Validated records automatically create or update Salesforce Opportunities and Engagement objects, populate CLM contract headers with obligation registers and renewal alerts, and attach source documents—all with full audit trails.

Step 7: Trigger downstream approvals and alerts

Cassidy initiates approval workflows, sets obligation due dates with owner assignments, schedules renewal reminders, and surfaces the new engagement data in portfolio dashboards for real-time visibility.