Agentic CX for large-operator scale
Telecom CX is high-volume, low-margin work. Conversations in many languages, often mixed mid-sentence; thin margin per ARPU; unforgiving unit economics. The Vihaya Engine’s CX reference solution is designed to lead with measurable touchless-resolution rate, not Net Promoter score theatre. India-operator context covered. Pre-revenue; no paid pilot has completed.
Workflows in scope
Multilingual intent routing
Classify the customer’s intent in any of 22 languages; route to the right resolution path with the agent’s recommendation attached.
Billing-dispute review
Read complaint + bill + plan + usage logs → cited refund / deny / escalate. CSR sees the rec in their existing console.
Churn / retention intervention
Score churn risk against historical patterns; recommend a grounded retention offer with cited rationale.
Plan-change consultation
Read the customer’s usage and recommend the plan that minimises spend — with the math shown.
Telecom FAQ
Which Indian languages does Vihaya CX support?
Strong day-one support for English, Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam, Punjabi, Odia, Assamese, and Urdu. The other scheduled Indian languages work but with thinner training data. Mixed-language conversations (English-Hindi code-switch) work natively because the underlying foundation models are trained on real Indian text.
Can this handle large-operator volume?
The engine is designed to scale horizontally. At large-operator scale (by published operator figures, tens of millions of interactions per day), the bottlenecks we’d expect are foundation-model rate limits and database read throughput — both addressable via provisioned capacity (Azure OpenAI India, Vertex Mumbai, or the customer’s region of choice) and standard scaling patterns. No production load test at that scale has happened. The pilot is scoped to one channel (one circle, one product) before extending.
How does the agent handle billing disputes?
The agent reads the customer’s complaint (call transcript, chat, email), the disputed bill, the rate plan, and the usage logs. It produces a cited recommendation: refund / partial / deny / escalate. The customer-care agent sees the recommendation in their existing console; the audit row preserves the decision trail for TRAI compliance and internal QA.
What about TRAI compliance and customer-data regulations?
TRAI’s customer-data norms and the DPDP Act both apply. Vihaya’s audit trail is designed to record every action with purpose and outcome; PII redaction in logs is configurable. CDRs (call data records) would stay inside the operator’s environment — Vihaya is designed to deploy in the operator’s VPC and does not egress data.
Want to see this in your environment?
30-minute discovery call. We follow up with a draft SOW shortly after.
Talk to us about a pilot