Session Objective

Keep moving the AI-inside-Litify track from package discovery into a practical implementation plan, with PHI controls appropriate for a personal injury firm.

What Was Accomplished

Created local blueprint: /Users/samaguiar/Documents/Projects/Repos/sail-litify-hub-mdt/Litify_AI_Integration_Project/PHI_SAFE_VERTEX_LITIFY_BLUEPRINT.md.

Live-validated the current org shape in LITIFY_ORG:

• litify__AI_API_Settings__c has 1 record.
• litify__AI_Document_Config__c has 0 records.
• litify__AI_Usage__c has 0 records.
• Matter Drive fields are live: Case_Folder_Google_Drive__c and View_Case_Folder__c.
• Litify Docs Matter linkage is live through litify_docs__Related_To__c and litify_docs__Related_To_Api_Name__c.
• Sampled Matter-linked PDFs have litify_docs__AI_Content__c = false.
• Damage, Note, and Intake fields provide usable anchors for medical bills, notes, and attribution workflows.

Updated local continuity logs:

• docs/litify-environment-diary.md
• OPS_LEDGER.md

Key Decision

Recommended a first-party architecture rather than depending on the dormant Litify AI package: Matter-page LWC, Salesforce staging objects, SAIL-controlled Google Cloud service, Vertex AI/Gemini for document analysis, and human-reviewed field suggestions.

PHI Guardrails

Do not route medical records, bills, insurance correspondence, file notes, or call summaries through generic gateways, Pipedream/MCP brokers, Perplexity, Semrush, marketing tools, or browser automation.

Before production PHI, confirm Google Cloud BAA, covered services, IAM, region, logging, retention, and least-privilege service account posture.