Client
Back to projects

automation

AI Outbound Call Platform

AI Outbound Call Platform is a system design portfolio project for Indonesian payment reminders and soft collection workflows. It focuses on controlled outbound calling where every attempt is checked against contact permissions, quiet hours, attempt limits, disputes, opt-outs, and suppression state before the dialer can call.

The design covers a Bahasa Indonesia AI voice agent using the OpenAI Realtime API, customer/account context, campaign orchestration, telephony provider routing, transcript and recording storage, outcome capture, promise-to-pay flows, payment-link triggers, compliance review, and human escalation for disputes or sensitive conversations.

Screenshots previewScreenshots

Feature walkthrough

Proof of what this project can do

Every entry pairs a recruiter-readable summary with screenshot evidence, build notes, and implementation context.

Features5
UpdatedJun 15, 2026
Page1 of 5
Tags

Feature 1 of 5

Compliance eligibility engine

Approves each outbound attempt only after checking consent or lawful basis, contact permission, quiet hours, attempt limits, active disputes, opt-outs, and suppression status.

Relevansi layanan

Contoh portfolio untuk keyword: jasa AI outbound call automation.

Studi kasus ini menunjukkan bentuk sistem yang bisa dibangun Arsitekode untuk client Indonesia yang membutuhkan jasa ai / whatsapp automation.

Lihat Jasa AI / WhatsApp Automation

AI voice call

payment reminder

soft collection

transcript review

Build Notes

01

MVP scope

  • One controlled payment-reminder campaign
  • One telephony provider
  • One Bahasa Indonesia AI voice agent
  • Customer import from CSV or admin form
  • Compliance checks before every call
  • Suppression list and opt-out handling
  • Transcript, outcome, and recording storage when legally permitted
  • Basic dashboard for review and manual QA

02

Recommended architecture

  • NestJS TypeScript backend with customers, accounts, campaigns, compliance, suppression, dialer, voice-agent, call-results, audit, and dashboard modules
  • PostgreSQL stores customer, account, campaign, outcome, promise, transcript, and audit data
  • Redis and BullMQ handle call jobs, retry state, and queue coordination
  • OpenAI Realtime API powers low-latency voice conversations
  • Twilio or a managed SIP provider supports the first pilot before deeper SIP routing work

03

Compliance boundary

  • Multiple outbound numbers are allowed for capacity, regional presence, failover, and campaign separation
  • Multiple numbers must not bypass a customer block, opt-out, dispute state, complaint, or attempt limit
  • The dialer only places calls after the compliance engine returns an approval for that specific attempt
  • Human escalation is required for disputes, sensitive cases, failed verification, complaints, and risky conversations

Tech Stack

Frontend

Next.js dashboardCampaign workspaceLive call monitorTranscript reviewCompliance queueAnalytics cards

Backend

NestJS modulesPostgreSQL modelsRedis/BullMQ jobsCompliance engineSuppression registryDialer integrationAudit logs

Screenshot Gallery

WhatsApp