im-calvin/cosona

COSONA 🦜

Inspiration

Our project, CoSona, was born out of a desire to humanize large language models (LLMs), which we observed had become increasingly robotic and devoid of unique perspectives. We, Kelvin, Jacky, Kevin, Can, and Ganesh, sought to create a chatbot that could be tailored to any specific use case, capable of impersonating any character, figure, or model. We envisioned a platform where users could upload content to construct a unique persona for their tasks, thereby personalizing their interaction with the chatbot. We had a vision of a chatbot that could assist users in learning a new language from their favorite TV characters or superheroes. We imagined a platform where users could engage in conversations with their favorite characters, getting to know them on a personal level, and even receiving the latest news report from a trending politician. This innovative approach to chatbot design was aimed at making the learning process more engaging and enjoyable for users. We also designed CoSona with a broader social goal in mind. We recognized the rising global issue of loneliness and saw an opportunity to address this through our chatbot. By creating a platform that allowed users to interact with their favorite characters in a fun, safe, and engaging environment, we hoped to provide a form of companionship and entertainment that could help alleviate feelings of loneliness. The potential applications of CoSona are vast. It could be licensed by schools as a tool to help combat depression and anxiety from a young age. Companies could use it to teach new languages in an engaging way. Call centers could use it to make their services feel more accessible and personal. By providing a unique and tailored experience, CoSona aims to increase acceptance of chatbots and make them a more integral part of our daily lives.

What It Does

CoSona aims to increase acceptance of chatbots, address the rising global issue of loneliness, and dive into the entertainment industry. It provides a safe and fun environment for individuals to chat. It can be licensed by schools to help combat depression and anxiety from a young age, by companies who want to teach new languages in an engaging method, and by call centers to make them feel more accessible and personal.

Key features:

• Tailored chatbot for any specific use case.
• Impersonates any character, figure, or model.
• Allows users to upload content to construct a unique persona for their task.
• Increases acceptance of chatbots and address global loneliness.
• Targets the entertainment industry, including discontinued series.
• Facilitates personal chats with favorite characters to learn about their opinions.
• Offers potential licensing opportunities for schools, companies, and call centers.

How We Built it

Front-end / Design

The front-end of CoSona was designed using Figma, a popular tool for interface design. The actual implementation of the design was done using NextJS, a React framework for building JavaScript applications. TailWind CSS was used for styling the application, providing utility-first CSS classes for rapid UI development. TypeScript, a statically typed superset of JavaScript, was used to ensure type safety and improve the development experience.

Backend

The backend of CoSona was built using Python. Cohere Coral, a language model, was used to power the chatbot functionality. Jupyter Notebook, an open-source web application, was used for creating and sharing documents that contain live code, equations, visualizations, and narrative text.

Future Viability

CoSona's future viability is promising due to its unique approach to chatbot technology and its potential impact on various industries. By impersonating any character, figure, or model, CoSona provides a tailored experience that can be used in a wide range of applications, from education to entertainment. The chatbot's ability to increase acceptance of this technology, combat global loneliness, and provide a more child-friendly option presents a significant market opportunity.

Moreover, CoSona's potential to be licensed by schools, companies, and call centers further expands its reach and potential for revenue generation. The planned expansion to allow users to create their own characters from scratch directly within the platform also indicates a commitment to continuous growth and adaptation to user needs.

Given the rising trend of chatbot usage and the increasing demand for personalized and engaging methods of interaction, CoSona is well-positioned for future success.

Challenges We Ran Into

Our team, consisting of Kelvin, Jacky, Kevin, Can, and Ganesh, faced several challenges while building the application due to our geographical distribution across three different time zones. Can was based in the UK, while the rest of us were in Canada.

All of our technical issues with LangChain was handled by Mentors. Thanks a lot to all the mentors that helped us.

Accomplishments That We Are Proud Of

We have developed a chatbot that is not only tailored to specific use cases but also capable of impersonating any character, figure, or model. This unique approach allows users to engage in meaningful conversations with their favorite characters, learn new languages in an engaging way, or even receive news updates from trending figures. Our chatbot is designed to combat the rising global issue of loneliness and to provide a safe, fun, and personal environment for users. We have also made strides in the education and entertainment sectors, and we are continuously working on expanding our services to allow users to create their own characters from scratch. The potential impact of CoSona on industries such as entertainment, counseling, and education is immense, and we are proud to be part of Cohere Coral Hackathon.

What We Learned

Most members of our team learned how to build applications using the Cohere API, which offers cutting-edge language processing capabilities. We benefited greatly from the help of mentors while working on LangChain. The support and guidance from experienced professionals in the field were invaluable in helping us develop our skills and create CoSona using the Cohere API and LangChain framework.

What's Next for CoSona

CoSona has a promising future with plans to expand its services and impact various industries. The next steps for CoSona include:

1. Expanding to allow users to create their own characters from scratch directly within the platform. This feature will provide users with even more customization options and enhance the overall user experience.
2. Influencing the chatbot market by presenting a more child-friendly option. This will help CoSona appeal to a broader audience and increase its market share.
3. Exploring potential licensing opportunities for schools, companies, and call centers. This will help CoSona generate revenue and expand its reach across different sectors.
4. Continuing to improve and refine the chatbot's capabilities, ensuring it remains at the forefront of chatbot technology and meets the evolving needs of users. By focusing on these goals, CoSona aims to solidify its position as a leading chatbot solution and continue to make a positive impact on the entertainment, counseling, and education industries.

Build With

Front-end / Design

● 🎨 Figma ● 📄 NextJS ● 💨 TailWind CSS ● ⌨ TypeScript

Backend

● 🐍 Python ● 🪸 Cohere Coral ● Jupyter Notebook

Set Up Instructions

Frontend

npm run dev - run in dev mode
npm run build - build the project
npm run start - start the prod ver of the frontend

Backend:

python3 api.py - run the api server in dev mode
docker pull imcalvin/cosona:latest - pull the docker image from docker hub and run that locally, alternatively:
docker-compose up