home / work / sayspark

EDTECH — NATURAL-LANGUAGE ROBOTICS

SaySpark

Robotics education through natural language. Kids tell the robot what to do in plain words — and it does it. No syntax errors, no barriers to entry.

LLMsSpeech InterfacesRoboticsEdTechProduct Design
HERO PHOTO: SaySpark robot with a student

## Problem

After years of mentoring hundreds of elementary students in robotics across Windsor, Ontario, one pattern was undeniable: kids love robots, but syntax-heavy programming stops many of them before the fun starts. A missing semicolon shouldn't end a child's engineering journey.

## Solution

SaySpark lets children program robots through natural language — spoken or typed. An LLM-powered layer translates a child's intent into safe robot behaviors, so the first experience of robotics is creation, not frustration. It's the foundation of my current venture, Aibotics.

## Demo

VIDEO: child commanding the robot in plain English (embed here)

## Features

  • Natural-language robot programming (speech and text)
  • LLM-powered intent → behavior translation with safety limits
  • Progression path from natural language toward real code
  • Classroom-ready hardware design

## Tech Stack

LLM APIsPythonEmbedded C++Speech RecognitionCustom Hardware

## Challenges

  • Constraining open-ended language into safe, predictable robot actions
  • Designing for young users: durability, simplicity, delight

## Lessons Learned

  • The best product insights came from the classroom, not the workbench

## Gallery

PHOTO: prototype
PHOTO: classroom testing
SCREENSHOT: software interface