• Retail & E-Commerce
  • India
  • Updated

Terminal Support by Voice for Pine Labs Merchants

Merchants who run their counters on Pine Labs terminals call for help with settlements, terminal set-up, connectivity and printing, often with a customer still waiting. We built a voice assistant that answers in the merchant’s language, follows Pine Labs’ own support procedures, runs the routine checks itself and hands over to an agent with the full context when a person is needed.

Book a 30-Minute Working Session
A card payment terminal and a receipt roll on the counter of a neighbourhood store at dusk

The situation

Every minute on hold is a minute without payments.

Pine Labs merchants run their counters on its POS terminals, and when a terminal misbehaves the merchant calls for help, usually with a customer still waiting. Most calls are the same handful of questions: a settlement that has not arrived, a terminal to set up or reactivate, a connectivity drop, a paper roll or printing fault, a card or EMI transaction to check. Merchants call in several Indian languages, at all hours, and peak trading days bring peaks in calls. Every minute on hold is a minute the counter cannot take payments.

What we built

A voice assistant that follows Pine Labs’ own procedures.

We built a voice assistant that picks up the merchant's call, identifies the merchant and the terminal, and works the problem the way Pine Labs' own support procedures say it should be worked. Azure Speech handles recognition and voice in the merchant's language. Agents built on Azure AI Foundry with Azure OpenAI understand the request, look up the terminal and the transaction, and walk the merchant through the fix step by step. Answers are grounded in Pine Labs' support procedures through Azure AI Search, so the assistant never improvises a policy. When a call needs a person, it passes to an agent with the transcript, the checks already run and the merchant's details, so nobody has to repeat themselves. Every call is logged for quality review and for the operations dashboards.

How it works

The architecture, stage by stage.

  1. Listen & speak

    • Azure Speech, recognition and voice in the merchant's language
  2. Understand

    • Azure AI Foundry agents on Azure OpenAI, intent, merchant and terminal identified
  3. Resolve

    • Guided fixes grounded in Pine Labs' support procedures through Azure AI Search, settlement and transaction checks
  4. Hand over

    • Escalation to a support agent with the transcript, the checks run and the merchant's details
  5. Learn

    • Call reasons and outcomes in Power BI for quality review

What changed

What the team has now.

  • Routine terminal questions are answered by voice at any hour, without a queue.
  • Merchants are helped in their own language, following Pine Labs' own support procedures.
  • Calls that need a person reach one with the full context, the checks already run and nothing to repeat.
  • Support leads see call reasons and outcomes in one place, so recurring terminal faults surface early.

Facing the same problem?
Start with a working session.

Bring your version of it to a 30-minute working session. We will walk through how this was built and what it would take on your data.