Teltruva

• 

  Describe the issue and upload supporting documents such as receipts, confirmations, or screenshots.

• 

  Teltruva extracts key evidence and organizes it into a structured case for the user to review.

• 

  Identify an official contact channel and use the completed inquiry to take action.

• 

  Review the one-time support package and complete a test payment through PayPal Sandbox.

• 

  Review and edit an evidence-grounded inquiry before translating it into the recipient’s language.

• 

  Start in your own language and turn a frustrating consumer issue into clear action.

Inspiration

Language barriers become especially serious when people need to resolve a problem involving a hotel, purchase, reservation, delivery, or other service. A user may have receipts, confirmations, screenshots, and other supporting evidence, but still struggle to explain what happened, determine what information matters, and contact the right organization in another language.

Teltruva was created to turn that frustrating process into a clear path from evidence to action.

What it does

Teltruva helps users create clear, evidence-grounded inquiries across language barriers.

A user explains the issue in their preferred language and uploads supporting documents, such as receipts, booking confirmations, or screenshots. Teltruva extracts relevant evidence, organizes the information into a structured case, identifies missing details, and generates an editable inquiry.

The user can review and revise the inquiry before translating it into the recipient's language. Teltruva also helps identify an official contact channel without guessing or automatically confirming unverified contact information.

For the PayPal AI Hackathon, Teltruva includes a one-time purchase flow powered by PayPal. The user can review what is included, complete the payment step, and continue to AI-assisted case support.

How we built it

Teltruva is a web application built with React, TypeScript, and Vite.

NVIDIA OCR extracts relevant information from uploaded documents. NVIDIA Nemotron, accessed through Nebius Token Factory, serves as the reasoning layer that organizes the evidence into a structured case and generates a grounded inquiry.

A web search API helps identify an official contact channel. Teltruva presents the result as a candidate for the user to verify instead of guessing or automatically confirming contact information.

PayPal Sandbox is integrated into the one-time purchase flow. The Sandbox environment allows the full payment experience to be demonstrated without processing a real charge.

Challenges we ran into

One of the main challenges was balancing automation with user control. Teltruva must reduce manual effort without making unsupported assumptions about the user's case or the recipient's contact information.

Another challenge was turning unstructured information from receipts, screenshots, and user descriptions into a consistent case structure that could support accurate inquiry generation and translation.

We also needed to integrate PayPal into the user journey without distracting from the core experience. The payment screen therefore explains what the user receives before the user proceeds to payment and returns to the case-support workflow.

Accomplishments that we're proud of

We created an end-to-end working prototype that connects document evidence, AI reasoning, multilingual inquiry generation, translation, official contact discovery, and a PayPal Sandbox payment flow in one guided experience.

We are especially proud that Teltruva does more than directly translate text. It first helps the user organize the case, identify relevant evidence, review the generated inquiry, and remain in control before taking action.

What we learned

We learned that supporting users across language barriers requires more than translation. The system must understand the user's intent, organize the evidence, preserve important facts, and communicate the issue clearly without changing its meaning.

We also learned that a payment integration should be part of a clear value exchange. The user should understand what is included before proceeding to payment.

What's next for Teltruva

Our next steps include expanding language support, improving document extraction across different receipt and confirmation formats, and refining official contact discovery.

We also plan to strengthen authentication, saved-case management, payment confirmation handling, and production-ready privacy and security controls.

Our long-term goal is to make Teltruva a trusted multilingual support tool that can help users prepare inquiries for email, official web forms, and chat-based contact channels.

Built With

• javascript
• nebius
• nvidia
• ocr
• paypal
• react
• sandbox
• token
• typescript
• vite