Session Summary

I reviewed the live Salesforce/Litify org, Notion history, local repo notes, and current Salesforce/Litify public capability references for an agent-driven case management build. No live Salesforce changes were deployed. This was a read-only feasibility and build-readiness pass.

What Was Verified

• Salesforce CLI is connected to LITIFY_ORG for sam@kylawoffice.com.
• Installed packages include Litify, Litify Docs/Docrio, Docrio SDK, Litify AI Data Extraction, Litify PM Agentforce, RingCentral, SMS-Magic Converse, Case Status, Clientell AI, and supporting Litify packages.
• Agentforce Studio and Agentforce Grid apps are present.
• Two active Agentforce employee agents exist: Litify Conflict Check Agent and Litify Matter Summary Agent.
• Sam has Agentforce User, Agentforce Agent Sam_s_Agent Permissions, Litify AI Admin, Litify AI Ask A Document User, and Docrio Email and Calendar End User permission sets.
• Current Agentforce actions are mostly conflict-check and matter-summary actions. No build-specific actions yet exist for LOR drafts, PIP logs, dec page follow-up, provider extraction, medical bill damage staging, or birthday SMS.
• Docrio templates already exist for MVA KY letter of representation to PIP carrier, BI carrier, PIP log request, health insurance provider LOR, disability insurance provider LOR, County Clerk Insurance Request, and related letters. The key templates are not currently marked as invocable templates.
• Request object supports the needed request types, including Insurance LOA, PIP Log, Medical Records, Medical Bill, Records & Bills, Insurance Policy, County Clerk: Insurance, and Police Report.
• Damage object supports the needed fields for medical bill staging, including Matter, Provider, Amount Billed, Amount Paid, Amount Due, Service Start/End Date, Type, Request lookup, and External Id.
• RingCentral/RingSense flow is active: RingCentral_Process_RingSense_Calls, record-triggered after Task creation. Live counts showed 14,330 RingSense Tasks, 4,364 in the last 30 days, but only 238 marked processed.
• Birthday/email-related automation exists: Check_Receive_Birthday_Emails_for_Clients is active. Account has birthday, phone, Receive_Birthday_Emails__c, and SMS_STOP__c fields. Matter also has SMS_STOP__c.
• Docrio document volume is high: 214,804 litify_docs__File_Info__c records, 3,176 created in the last 30 days. Last 30 days included 237 medical bill files, 287 medical record files, 198 letter-of-representation files, and 1 PIP log file by filename. All last-30-day Docrio file category values were null, so classification hygiene is a real blocker for clean document AI triggers.
• Litify AI Data Extraction is installed but its AI document config, object config, and field config tables had 0 records. Ask A Document had only one police-report record. This means AI document extraction is available but not configured for the workflows requested.

Feasibility Read

The work is feasible, but the right posture is agent-assisted, not autopilot. Salesforce should detect events and generate drafts or candidate records. Staff should review and approve before letters are sent, text messages are sent, requests are ordered, or damages are committed.

The strongest immediate build is a Litify Agent Review Queue action layer. It can accept candidate actions from flows, RingSense parsing, Docrio upload classification, scheduled checks, or Agentforce. Each candidate should include source record, extracted facts, confidence, proposed action, owner, due date, approve/reject controls, and an audit trail.

Best first lanes: