Sixteen-language understanding
One assistant that reads, reasons and replies across sixteen Indian languages and their native scripts.
Agentic AI
Sixteen scripts.
One assistant.
Swadesh is Chat360's indigenous large language model for Indian government and enterprise. It understands sixteen Indian languages in their native scripts and acts across the channels and systems citizens and customers already use.
Swadesh · sample conversation
हिन्दीमुझे पीएम किसान का पैसा नहीं मिला है
अगर e-KYC अधूरी है, तो किस्त रुक सकती है। मैं e-KYC पूरी करने के चरण समझा सकता हूँ।
தமிழ்என் புகார் நிலை என்ன?
உங்கள் புகார் எண்ணைச் சொன்னால், நிலையைப் பார்க்கும் வழியை நான் விளக்குகிறேன்.
Languages
One assistant across sixteen Indian languages
Each language is supported in its native script, so customers write the way they speak and read replies that sound native — not translated. Built for the multilingual access priorities behind Digital India and Bhashini.
- हिन्दीHindiDevanagari
- বাংলাBengaliBengali
- मराठीMarathiDevanagari
- తెలుగుTeluguTelugu
- தமிழ்TamilTamil
- ગુજરાતીGujaratiGujarati
- اُردُوUrduNastaliq
- ಕನ್ನಡKannadaKannada
- ଓଡ଼ିଆOdiaOdia
- മലയാളംMalayalamMalayalam
- ਪੰਜਾਬੀPunjabiGurmukhi
- অসমীয়াAssameseAssamese
- मैथिलीMaithiliDevanagari
- कोंकणीKonkaniDevanagari
- नेपालीNepaliDevanagari
- संस्कृतम्SanskritDevanagari
16 Indian languages shown here.
The model
An indigenous LLM, not an import
Swadesh is built in India for Indian language intelligence, aligned with the multilingual access priorities behind Digital India and the national strategy for AI. Indian scripts, code-mixed speech and regional phrasing are handled as first-class inputs rather than edge cases handled after an English translation step. Swadesh is built and available for department-specific deployment on the Chat360 platform, subject to configuration and environment requirements.
Native script comprehension
Devanagari, Bengali, Tamil, Telugu, Kannada, Malayalam, Gurmukhi, Odia, Gujarati and Nastaliq are read directly.
Code-mixed conversation
Customers who switch between English and their language mid-sentence stay understood.
Departmental and enterprise context
Responses are grounded in approved knowledge sources and connected systems, not in open-ended generation.
Hosted in India
SwadeshGPT supports India-hosted deployment on MeitY-empanelled cloud infrastructure. Depending on the approved deployment architecture, inference and application data can remain within India-hosted or customer-controlled environments.
Platform
Full-stack: channels, intelligence and government systems
Swadesh runs inside the Chat360 Agentic AI platform, connecting the channels customers use to the systems your teams already run on.
Layer 01
Conversational channels
- Google RCS
- Voice AI
- Web chat
Layer 02
Swadesh intelligence layer
- Sixteen-language understanding
- Intent and context
- Agentic reasoning
Layer 03
Government & enterprise systems
- CRM and departmental databases
- Grievance and case management systems
- Scheme registries and knowledge bases
- Approved identity and document services
- Analytics
Capabilities
Everything a citizen-facing assistant needs
A complete operating layer for multilingual conversations — delivery, reasoning, integration, governance and measurement.
Omnichannel delivery
Deploy the same assistant across WhatsApp, Google RCS, Voice, web chat and email from a single configuration.
Agentic workflows
Move beyond answers: Swadesh can take the next step in a journey and act inside connected departmental and enterprise systems.
Government & enterprise systems integration
Connect departmental databases, knowledge bases and business applications so conversations run on real, current data.
Governance and controls
Role-based access, approval flows and auditable conversation history built for departmental operating standards.
Analytics and measurement
Track containment, resolution and conversion by language and channel so impact is measurable, not anecdotal.
See what Swadesh can automate for your department.
Use cases
Where Swadesh serves citizens and departments
Eight high-volume conversation types across citizen services and departmental operations.
- G2C
Citizen service desk & scheme discovery
Help citizens discover central and state schemes — PM-KISAN, Ayushman Bharat, PM Awas Yojana — and understand published eligibility criteria, by voice or text in their own script.
- G2C
Grievance redressal & support
Lodge and track citizen grievances over WhatsApp, Google RCS or IVR voice, with status updates returned in the language the complaint was filed in. Can be configured to connect with departmental grievance systems such as CPGRAMS through approved APIs, where access is available.
- G2C
Citizen onboarding & verification
Guide applicants through document requirements, e-KYC steps and application walkthroughs for passports, certificates and ration cards.
- G2C
Disaster management & emergency alerts
Deliver broadcast communications and voice guidance during emergencies and weather alerts, in regional languages and dialects.
- G2C
Application & service status
Send proactive status updates on applications and services, and handle the follow-up questions those updates generate without adding ticket volume.
- G2G
Field & Anganwadi / ASHA enablement
A multilingual voice assistant to help frontline workers access manuals, report field data and interpret guidelines in real time, on the devices they already carry.
- G2C / G2B
Tax & utility payment guidance
Assist with GST queries, municipal and property tax, and utility bill extensions, pointing citizens and businesses to the correct process and department.
- G2G
Departmental knowledge search
Answer questions from approved circulars, guidelines and internal knowledge sources, with the response returned in the language the question was asked in.
Ready to build multilingual citizen and customer conversations at government and enterprise scale?
Why Swadesh
Indian language intelligence, built for government workflows
Built for Indian language, not translated into it
Swadesh is trained on Indian languages and scripts as first-class inputs, rather than routing everything through an English intermediate step.
Built on a platform already in production
Swadesh runs on the Chat360 Agentic AI platform, with the same governance, access controls and operating model already used across enterprise deployments.
Agentic, not just conversational
Swadesh is designed to complete work inside connected systems, so a conversation ends in an outcome rather than a handoff.
Get started
Build multilingual citizen services at scale
See what Swadesh can automate across your channels, languages and systems.