Telco / Digital Economy
Malaysian telco automates interconnect settlement across 13 carriers
CLIENT
A leading Malaysian telecommunications operator
Cadence
Monthly cycle
Client Overview
A Malaysian telecommunications operator settling interconnect payments each month with thirteen domestic voice and SMS carriers. Three people worked a roughly 132-step sequence, building payment-request files and chasing approval status by hand.
How the Workflow Runs
The monthly interconnect-settlement cycle across 13 carriers now runs through a single automated pipeline. The request is validated before it is submitted, and every approver mismatch or missing reference file is routed down a defined exception path.
FIG. 01 · Monthly interconnect settlement across 13 carriers
01
Build & submit the request
Bot
On a fixed monthly schedule, the bot builds the payment-request file for the period’s carrier invoices from a maintained reference table and uploads it into the approval workflow.
02
Validate approver & reference data
Agent
The process checks the invoice’s named approver against the maintained approver list and confirms carrier reference data is complete. Any gap is flagged as a business exception, not pushed through.
03
Human checkpoint
Human
Users are emailed any flagged approver or reference mismatch, and any login failure. They correct the source file for the next run or refresh credentials through a self-service console. The bot halts rather than retrying blindly.
04
Track status & distribute advice
Bot
Queue-triggered processes extract live payment status and compile a status report, then pull the payment advice document from the core financial system and email it to the business users.
How Cognitive Consulting helped
- 132 steps across 13 carriers, once a month, under a deadline, by three people. We collapsed it into one pipeline and ended the status-chasing.
- Validation moved to the front. A wrong approver code now surfaces at build time instead of weeks later in a reconciliation.
- On a failed login the bot halts. Retrying blindly is how automations get service accounts locked, so credential refresh goes back to the business.
OUTCOMES
- Approver mismatches and incomplete reference data are flagged as business exceptions, never pushed through
- Credential failures halt the bot rather than retry blindly, keeping access control with the business
- Status tracking and payment-advice extraction each carry their own audit trail
RELATED CASE STUDIES