Inspiration
Since the launch of GPT-3 by OpenAI in 2020, people have primarily used AI for information seeking, content creation, and personal productivity. Initially, these tools were perceived as a breakthrough in innovation for its fast responses for complex questions through simple interfaces, and since then it rapidly became a seamless extension of human capabilities.
But, it didn’t take long before the consequences began to emerge. UNESCO has warned that the continued reliance on AI in education compromises critical thinking and independent learning, not only within school but beyond daily life. As students and self-learners increasingly turn to AI for answers, people tend to use AI for everything without carefully reviewing what the model is providing.
What it does
Cliro was born out of the necessity to learn with current tools but without relying on them to think for us. Cliro is not like any other AI that doesn't care about the user understanding. Cliro is a writing companion, designed to encourage creativity by helping the user only after he already starts thinking, when words are hard to find. With a simple click on any selected text (whether you wrote it or not), Cliro provides immediate assistance to: summarize, explain, rewrite and translate. And that's how Cliro not only encourages you to read or write, but to analyze beforehand, so the user can know specifically where does he need help.
It provides structure and clarity, overcoming writer's block and making your writing faster, clearer, and more effective, anytime and anywhere.
How we built it
For this MVP, we decided to develop Cliro as a Google Extension that will work in the browser to help user with daily simple tasks. Cliro works by using the Gemini API directly in our Fast API backend in order to generate responses to the user requests. Our frontend was developed using React JS and a Manifest JSON that will help us set up our environment for Google Chrome usage.
Challenges we ran into
The biggest challenge out team ran into is when our tool had to read and modify the DOM the most efficient and global way. Because not only some websites had limits when trying to modify and extracting data, but also to adapt the tool to work with different website restrictions and limits.
Accomplishments that we're proud of
Fortunately, Cliro is now fully operational and ready to deploy. The team is excited to continue developing the project and motivation goes beyond offering daily assistance to users. This tool is designed to maximize self-learning and encourage creativity by making every day, rapid requests easier to handle by making Cliro a more intuitive alternative. By making it easier to follow a self-made idea path rather than a fully AI driven idea through LLMs chatbots.
What we learned.
New technological tools help us optimize many of our daily tasks, but certain activities still require human intervention. This isn't because technology can't handle them, but because we're concerned about our health and our future as professionals. As a startup, we discovered and learned to collaborate to develop solutions that fulfill our commitment to reducing reliance of students and professionals on purely AI-generated responses.
What's next for Cliro
Google Chrome Extension was our first step with this project. However, we plan to focus the tool on helping the user learn and providing them with various study methods and sources through:
- A proprietary AI model with NLP and Generative AI
- Specialized web-seeking agents that display videos and/or articles from the web (depending on user preferences)
- Adaptations for a B2B focus, either with specialized RAG models (depending on the client: schools, offices, code, etc.)
- Implementation of ML flows and Agent Tracing when seeking information
- Delivery of simple yet key responses, encouraging users to conduct their own research by learning about key terms
- And most importantly: A Desktop program functional across all applications, and even compatible with code
Log in or sign up for Devpost to join the conversation.