AI-Driven Case Triage & Resolution System

A Salesforce portfolio project that demonstrates an Agentforce-style support workflow: classify case intent and urgency, retrieve governed CRM context, apply trust checks, and produce agent-facing recommendations.

This repository has two parts:

• Working web demo: A Vite + React console that reviewers can use without Salesforce org access.
• Salesforce DX source: Apex, LWC, and Platform Event examples that show how the workflow maps into a Developer Edition org.

Live Demo

GitHub Pages deploys from main.

https://parvezsk3.github.io/ai-driven-case-triage-resolution/

Why This Project Exists

The resume project says:

Built an Agentforce case triage system to sort cases by intent and urgency. Added Einstein Trust Layer controls to protect sensitive customer data. Pulled CRM context from Data Cloud through REST APIs and showed case recommendations in a custom LWC. Cut average handling time by 24.6%.

This repo makes that project reviewable through a live demo, Salesforce source, and deployment notes.

Demo Features

• AI triage queue with realistic support cases.
• Intent, urgency, sentiment, confidence, and owner classification.
• Data Cloud-style context panel for entitlements, case history, and account signals.
• Protected signal panel for governance and trust review.
• Agent-facing recommendation with human-review guardrails.
• Step-by-step pipeline trace from classification to Lightning Web Component exposure.

Salesforce Source Included

Area	Source
Apex recommendation service	force-app/main/default/classes/AiCaseTriageService.cls
Apex tests	force-app/main/default/classes/AiCaseTriageServiceTest.cls
Triage audit event	force-app/main/default/objects/Ai_Triage_Outcome__e
Case recommendation LWC	force-app/main/default/lwc/aiCaseRecommendation

Local Demo Setup

npm install
npm run dev

Open the local Vite URL shown in the terminal.

Build and Quality Checks

npm run lint
npm run build

Salesforce Developer Edition Deployment

The Salesforce CLI is required for org deployment. It is not required for the public web demo.

sf org login web --alias ai-triage-dev --set-default
sf project deploy start --source-dir force-app --target-org ai-triage-dev
sf apex run test --target-org ai-triage-dev --class-names AiCaseTriageServiceTest --result-format human --code-coverage

After deployment:

1. Add AI Case Recommendation to the Case Lightning Record Page.
2. Create sample Case records with Access, Billing, Order Status, and Product Defect subjects.
3. Confirm the LWC displays intent, urgency, owner queue, confidence, and human-review status.
4. Review the Ai_Triage_Outcome__e Platform Event definition for audit and downstream workflow handoff.

Access Needed Later

To deploy this into a Developer Edition org, I will need:

• Salesforce Developer Edition org login or temporary access.
• Permission to install/use Salesforce CLI locally.
• Permission to deploy Apex classes, LWC, and Platform Event metadata.
• Confirmation of whether you want the next iteration connected to a real prompt/model provider or kept deterministic.

Tech Stack

• React 19
• TypeScript
• Vite
• Salesforce DX source format
• Apex
• Lightning Web Components
• Platform Events
• Agentforce-style workflow design
• Data Cloud-style context modeling
• GitHub Actions
• GitHub Pages

Portfolio Notes

This is a realistic personal project. It does not claim to use real customer data, live Data Cloud records, or a production model provider. The demo uses deterministic fixtures so recruiters and interviewers can understand the architecture and governance decisions without needing org access.