←Back/Product/Jev
IntermediateProduct

How to Build a Real-Time Voice-Controlled To-Do List App with Jev

Create a fluid, hands-free to-do list that updates instantly as you speak. Use a multi-step classification process to understand commands, match tasks, and execute functions in real-time without needing to pause.

Try this with your AI assistant

Download the skill for this workflow. Version 1.0.1.

Download skill
How to Build a Real-Time Voice-Controlled To-Do List App with Jev

Before you start

Tools used

  • Jev

    TypeSafe AI’s model for fast structured decisions, including choices, scores, and boolean probabilities.

    VisitJev

The workflow

Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.

5 steps

Step 01

Set Up Live Transcription

Continuously capture audio from the user's microphone and convert it into a live text stream. This stream will be the input for your decision model.

Step 02

Classify Action Intent

Feed the transcribed text stream to a Jev model. Configure it to analyze the incoming words and decide when a complete, actionable command has been stated, without requiring the user to pause.

This step is crucial for creating a fluid user experience. The model needs to determine the boundary of a command within a continuous stream of speech.

Step 03

Fuzzy Match the Task

Once Jev decides a command has been given, take the relevant text (e.g., 'oat milk') and use a second Jev call to perform a fuzzy match against the existing items in your to-do list. The model should return the best match and a confidence score.

Step 04

Match the Function

Simultaneously, use a third Jev call to identify the operation the user wants to perform (e.g., 'remove', 'complete', 'low priority'). Match this operation to a corresponding function in your application's code, like removeTask() or setPriority().

Example prompt
Book dentist appointment, remove. Buy oat milk, complete. Review pull request, low priority.
Step 05

Execute the Action

Using the matched task from Step 3 and the matched function from Step 4, execute the corresponding code to update the to-do list state. Update the UI after the action succeeds. End-to-end latency depends on transcription, model calls, and your application.

Build your next product with ChatPRD

Turn an idea into a PRD, user stories, and a plan.

Try ChatPRD free

Start shipping better products.

Join 250,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.

Free to start · 14-day Pro trial · SOC 2 Type II