askly

next.jsconvexopenaiturborepo

the AI-first communication layer for modern institutions. provides 24/7 automated student support in english, hindi, and regional languages, reducing repetitive queries by 70% with context-aware multi-turn conversations.

view source (github) demo →

Metrics

query reduction

70%

repetitive queries automated

languages

english, hindi, regional

availability

24/7

zero staff overhead

Architecture

Student (Chat/Voice)

Next.js Frontend

Convex Backend

AI Engine (OpenAI/Gemini/Claude)

Vapi (Voice)

Real-time Backend: Uses Convex as a serverless real-time database for instant message delivery and persistent conversation history.

AI Pipeline: Multi-model architecture supporting OpenAI, Google Gemini, and Anthropic Claude for context-aware multilingual responses.

Code Snippet

apps/web/src/app/chat/page.tsx

// embeddable widget integration
<script
src="https://askly.example.com/widget.js"
data-org-id="your-org-id"
data-theme="light"
data-lang="en"
/>

Production Readiness

Communication

Multilingual Support: Seamless conversations in English, Hindi, and regional languages with automatic translation.
Voice Integration: Vapi-powered voice calls for students who prefer speaking over typing.

Integration

Embeddable Widget: Single script tag deployment with organization-specific configuration.
Clerk Authentication: Secure multi-tenant access with role-based permissions for institutions.

About

Askly tackles the problem of campus offices handling thousands of repetitive queries daily through English-only systems. Built as a monorepo with Turborepo, it combines a Next.js 15 frontend with Convex real-time backend and multi-model AI to deliver instant, accurate responses in the student’s preferred language.