Voice help for public services.

Voice Seva helps people complete official service tasks through short spoken guidance, service memory, and the right next official route.

Open Voice Seva See service scope

Platform scope

Voice-led public-service guidance.

Core pillars

Voice-first interaction

Users begin by speaking naturally instead of decoding forms, menus, and portal structure.

One-next-step guidance

Every turn stays short, specific, and actionable so the user never gets a full procedure dump.

Context continuity

Qdrant-backed memory preserves the task, the current step, and repeat requests across the session.

Public-service guardrails

Service playbooks, official links, and handoff notes keep the assistant inside trustworthy public-service guidance.

One step

Public service

Speak once. Get the next route.

24/7

Platform focus

Built for high-friction public-service tasks.

The product works best when someone knows the task they need, but not the portal path, document order, or next official action. Voice Seva reduces that confusion before it turns into drop-off.

Users can start with one spoken request such as 'I need voter ID correction' or 'I want to update Aadhaar'.
The platform responds with one next step instead of a long form walkthrough.
Each answer stays tied to the public-service route rather than broad assistant chatter.

See societal impact

Conversation engine

One engine runs text and live voice.

Voice Seva uses Vapi for live voice, Qdrant for retrieval and memory, and a server-side turn engine that keeps the guidance deterministic even when the wording is rewritten for clarity.

Vapi handles microphone sessions, transcripts, and spoken replies.
Qdrant stores memory and curated public-service knowledge for the next turn.
The turn engine can repeat, simplify, and continue without resetting the service context.

Open voice experience

Voice engine

Memory

Qdrant keeps the task context.

Voice and text stay on one engine.

FAQ

Common questions about Voice Seva

These answers explain the live platform as it exists today: voice-first, public-service focused, and designed to guide the next official step rather than replace official systems.

Who is Voice Seva built for?

Voice Seva is built for people who struggle with form-heavy public-service flows, especially low-literacy users, elderly users, first-time smartphone users, and assisted-access contexts.

Can someone start without typing?

Yes. The primary experience is voice-first. Text exists only as a fallback so the flow can still be tested or used when voice is unavailable.

Does it remember the previous step?

Yes. The system stores the active service, the current step, and recent context so it can repeat or continue without starting over.

Which services does the platform focus on now?

The live platform focuses first on Aadhaar address help, Aadhaar mobile-update guidance, document help, and PAN correction, with the service architecture already expanding toward voter ID, passport, ration card, and pension support.

Does Voice Seva replace official portals?

No. Voice Seva helps users understand the route, the required documents, and the next official step. It still hands off to the correct portal, service centre, or official process when needed.

How is the platform built?

Voice Seva uses Vapi for live voice sessions, Qdrant for memory and retrieval, and an OpenRouter-backed rewrite layer on top of a deterministic service engine.