NamiGenAI Voice Agents that sound human. And scale like software.
Bilingual voice AI stack for Vietnamese and Japanese conversations, built for enterprise workflows.
Overview
NamiGen is NamiTech's in-house voice AI stack built for Vietnamese and Japanese, with models owned across the full voice interaction layer from acoustics to dialogue.
Instead of relying on separate third-party components, NamiGen combines bilingual speech models, a reasoning runtime, and a closed learning loop into one integrated AI system.
This ownership allows enterprise teams to tune the stack for real operating conditions, including Vietnamese telephony noise, Japanese honorifics, banking vocabulary, HR vocabulary, regional accents, and complex contact centre scenarios.
What it does
NamiGen automates customer-facing conversations across every channel and every use case.
Operated on NamiChain, NamiTech's proprietary knowledge and business-rules platform, NamiGen always works from your current products, policies, and processes. Every response is accurate, consistent, and compliant, because the knowledge and rules it operates on are owned and updated by your team.
End-to-end latency is below 1.5 seconds, with high intent recognition accuracy validated in production deployments.
Platform Capabilities
Bilingual Telephony-Grade ASR
Vietnamese and Japanese ASR trained on in-domain telephony audio, including 8kHz audio, packet loss, codec artefacts, banking vocabulary, HR vocabulary, and contact centre scenarios. Handles regional accents including North, Central, and South Vietnamese, as well as Kansai, Tohoku, and Tokyo Japanese.
Bilingual Neural TTS (OmniVoice)
Bilingual neural TTS covering the same accent groups and supporting prosody control for sensitive conversation flows such as collections disclosures and hardship language.
Reasoning Runtime
Combines intent routing, retrieval-augmented FAQ, slot-filling, and mode-switching. Maintains multi-turn state across mode transitions, helping the conversation continue without context loss.
Continuous Learning and Optimisation
NamiGen continuously learns from operational interactions and customer feedback to improve performance, accuracy, and conversation quality over time, while maintaining business oversight and governance controls.
Compliance Policy Enforcement at Runtime
Compliance is handled as a policy filter inside the runtime, not only as post-call review. Runtime constraints include FSA disclosure rules, kasuhara handling, APPI, PDPA, and labour-law guards.
Audit Substrate
Every call is replayable, every tool call is traced, and every escalation is bucketed by failure mode. Supports region-aware routing for strict data residency requirements.
Use Case
Sales
Grow revenue from every conversation. NamiGen automates product consultation, lead qualification, and prospect nurturing across multiple interaction channels, 24/7, always following your compliance-approved scripts.
Automatically consult, qualify, and nurture leads 24/7.
Draw on live product knowledge, pricing, and policy managed by your team on NamiChain.
Transfer seamlessly to a human advisor with full conversation context when deeper consultation is needed.
Customer Care
Expand customer care capacity 24/7. NamiGen handles inbound service requests, answers product and policy questions, and resolves common issues automatically across voice, chat, and digital channels.
Automatically receive, answer, and process support requests 24/7 using natural conversation.
Draw on current product documentation, policies, and processes managed by the business.
Transfer to a human agent with complete conversation context when specialist handling is needed.
Collections
Scale debt recovery without sacrificing the customer relationship. NamiGen automates early-stage collection outreach including reminder calls, payment confirmation, and commitment capture, at a volume and consistency no human team can match. Interactions are personalised based on customer profile, payment history, and days overdue.
Automatically run reminder campaigns, information verification, and payment commitment capture 24/7.
Personalise interaction content based on customer profile, payment history, and overdue stage.
Transfer to a human agent with full context for cases requiring negotiation or specialist handling.
Record interaction evidence clearly, retrievable on demand.
Trust Proof
End-to-End Model Ownership
NamiTech owns the voice AI stack from acoustics to dialogue, including bilingual ASR, bilingual TTS, reasoning runtime, and learning loop.
Built on Proprietary Corpora
NamiGen is trained on proprietary Vietnamese and Japanese telephony corpora, including banking, HR, and contact centre scenarios.
Production-Tuned, Not Lab-Only
The system has been tuned under real workloads, including SHIFT in Japan, banking pilots in Vietnam, and the Prudential AI Interview project.
Measured Performance
End-to-end latency below 1.5 seconds. High intent recognition accuracy validated in production deployments.
Versioned and Rollback-able Learning
Call signals are converted into operator-approved, versioned, rollback-able updates before release.
Incident Hardening
Operational incidents are converted into regression tests. Every change is replayed against historical calls before release, with AI judges grading outcomes and humans approving changes.
Industry Application
Finance and Banking
Automate onboarding, account queries, transaction support, and debt recovery.
Insurance
Automate product consultation, policy queries, renewal reminders, and post-sale service.
Retail
Automate product consultation, order support, and loyalty programme queries.
Enterprise Contact Centre and BPO
Scale inbound, outbound, and self-service across multiple clients and campaigns.
Automate customer conversations at scale with AI voice agents
Request a Demo