AI Voice Assistant for Form Filling
Turn form filling into a guided conversation. Elinext's Voice Assistant helps people understand each question, dictate answers, and check important details before they are saved. English and Italian voice guidance works alongside an editable form, keeping users in control from the first answer to the final review.
React
TypeScript
Node.js
OpenAI
About the project
The challenge: Filling out a form can mean navigating unfamiliar questions, typing repetitive details, and correcting small mistakes. A useful voice assistant needs to explain what each field asks for, handle spoken corrections, and make it clear which answers have actually been saved.
The product: Elinext developed a voice assistant that brings conversational support directly into the form. Users can speak in English or Italian, ask what a field means, revisit an answer, or switch to typing whenever they prefer. The interface stays in English while the conversation follows the selected language.
The approach: A React and TypeScript frontend combines an editable form with live voice guidance powered by OpenAI over WebRTC. A Node.js and Express backend manages voice sessions and AI requests. Shared validation rules check proposed changes, while details such as dates, email addresses, phone numbers, and addresses require explicit confirmation after dictation.
The experience: Users can follow their progress, correct a suggestion, recover a draft in the same browser tab, and review their details before completing the demo. The project demonstrates a reusable interaction pattern for guided form filling. The current demo works with a defined set of fields; applying this approach to another form involves configuring its questions and validation rules.
Built with React, TypeScript, Node.js, and OpenAI
About the project
The challenge: Filling out a form can mean navigating unfamiliar questions, typing repetitive details, and correcting small mistakes. A useful voice assistant needs to explain what each field asks for, handle spoken corrections, and make it clear which answers have actually been saved.
The product: Elinext developed a voice assistant that brings conversational support directly into the form. Users can speak in English or Italian, ask what a field means, revisit an answer, or switch to typing whenever they prefer. The interface stays in English while the conversation follows the selected language.
The approach: A React and TypeScript frontend combines an editable form with live voice guidance powered by OpenAI over WebRTC. A Node.js and Express backend manages voice sessions and AI requests. Shared validation rules check proposed changes, while details such as dates, email addresses, phone numbers, and addresses require explicit confirmation after dictation.
The experience: Users can follow their progress, correct a suggestion, recover a draft in the same browser tab, and review their details before completing the demo. The project demonstrates a reusable interaction pattern for guided form filling. The current demo works with a defined set of fields; applying this approach to another form involves configuring its questions and validation rules.
Built with React, TypeScript, Node.js, and OpenAI
AI Voice Assistant for Form Filling
Voice guidance in English and Italian, confirmed field updates, and a clear review step bring conversational support directly into the form.
ReactTypeScriptNode.jsOpenAIAI Voice Assistant for Form Filling
Voice guidance in English and Italian, confirmed field updates, and a clear review step bring conversational support directly into the form.
ReactTypeScriptNode.jsOpenAI
Features
Guided Voice Conversations
Users choose English or Italian before starting the microphone, then work through the form one question at a time. The assistant explains unfamiliar fields, listens to natural answers, and uses the current form to guide the next step. A visible conversation transcript and question list help users follow along.
Voice and manual input share the same form state. Users can dictate a specific field, revisit an earlier answer, or type directly into the form. Pause and end controls keep the conversation under their control, while automatic idle disconnection avoids leaving an unused voice session running.
Confirm Before Saving
An email address, date, phone number, or street address can be easy to mishear. The assistant presents these dictated details as pending suggestions and asks the user to confirm them by voice or on screen. Users can correct part of an answer or dictate it again before accepting the result.
Shared rules validate formats and permitted ranges before updates reach the form. Revised suggestions invalidate older confirmations, and manual edits take priority over outdated AI actions. This makes the distinction between a proposed answer and an accepted value explicit throughout the conversation.
Review With Confidence
A progress indicator shows which required fields are complete and which still need attention. Once the form is ready, users review their details and can return to any field. The assistant supports the process, while the final acknowledgment and completion stay in the user's hands.
A tab-scoped draft preserves form values and pending suggestions after a reload or connection failure, so users can resume from their current question. Completing the sample form produces a local JSON download, demonstrating a complete journey from spoken answers to reviewed, structured information.
Features
Guided Voice Conversations
Users choose English or Italian before starting the microphone, then work through the form one question at a time. The assistant explains unfamiliar fields, listens to natural answers, and uses the current form to guide the next step. A visible conversation transcript and question list help users follow along.
Voice and manual input share the same form state. Users can dictate a specific field, revisit an earlier answer, or type directly into the form. Pause and end controls keep the conversation under their control, while automatic idle disconnection avoids leaving an unused voice session running.
Confirm Before Saving
An email address, date, phone number, or street address can be easy to mishear. The assistant presents these dictated details as pending suggestions and asks the user to confirm them by voice or on screen. Users can correct part of an answer or dictate it again before accepting the result.
Shared rules validate formats and permitted ranges before updates reach the form. Revised suggestions invalidate older confirmations, and manual edits take priority over outdated AI actions. This makes the distinction between a proposed answer and an accepted value explicit throughout the conversation.
Review With Confidence
A progress indicator shows which required fields are complete and which still need attention. Once the form is ready, users review their details and can return to any field. The assistant supports the process, while the final acknowledgment and completion stay in the user's hands.
A tab-scoped draft preserves form values and pending suggestions after a reload or connection failure, so users can resume from their current question. Completing the sample form produces a local JSON download, demonstrating a complete journey from spoken answers to reviewed, structured information.
Contact Us
Anastasia Timoshenko
Regional Account Manager
1000+
Delivered projects
300+
Clients worldwide
700+
In-house developers
28+
Years in industry
5.0 Ratings