Inspiration
Our team sympathized strongly with the visually impaired and felt the need to bridge the gap for internet usage amongst the disabled populations. There are 44 million around the world that are legally declared blind and roughly 73% of disabled Internet users suffer barriers on a regular basis, and we felt like there had to be a solution.
What it does
SightLine is an agentic AI powered browser extension that allows the visually impaired to access the internet using voice commands. Once activated, the browser awaits user dialogue and navigates the web, narrating the process along the way. SightLine is able to execute tasks end-to-end for its users, marking a major development in accessibility technologies.
How we built it
We built SightLine as a Chrome extension for voice input and in-browser actions, plus a Node.js/Express backend for AI command processing. Speech is transcribed in the browser, sent to the backend for intent detection, and then executed on the active webpage through content scripts. We also integrated browser-use Web UI for agent-style automation and added text-to-speech feedback so users hear clear confirmation after each action.
Challenges we ran into
We found it difficult to connect to all of our many APIs, leading to issues with the speech-to-text conversion as well as the task execution process. The biggest challenge was probably getting SightLine to execute smarter actions rather than simple Regex based tasks on the internet.
Accomplishments that we're proud of
What we’re most proud of is building SightLine as a real accessibility-first product in the 24 hours that we had: a complete voice-to-action browser assistant that goes from spoken command to real webpage interaction with immediate spoken feedback.
We integrated a lot of systems into one pipeline, including Chrome Extension APIs, Web Speech recognition, text-to-speech, a Node/Express backend, WebUI, and a fallback command engine for reliability when APIs fail or keys are missing.
Beyond single commands, we enabled multi-step planning and optional agentic automation, showing a clear path from MVP voice control to more advanced autonomous browser workflows.
Most importantly, we built this with inclusive intent. SightLine is designed to reduce friction for people who benefit from hands-free and voice-led browsing, turning accessibility from an add-on into the core experience.
What we learned
We learned how to connect speech recognition, AI intent understanding, and browser automation into one reliable end-to-end system. We also learned that handling edge cases like permissions, restricted tabs, and script injection is just as important as building the core feature. In conclusion, we learned how to build fast as a team while prioritizing accessibility in our design decisions.
What's next for SightLine
Next for SightLine is always-on, wake-word activation (like Amazon Alexa or Google Home) so users can start commands hands-free. We also plan to expand command coverage so SightLine can handle more complex multi-step tasks across shopping, forms, productivity, and everyday browsing workflows.
Log in or sign up for Devpost to join the conversation.