OrgSage | Devpost

Inspiration

We saw a need for a unified solution to help organizations internally manage information across various platforms. Organizations like ACM use many tools like Notion, Discord, and Slack, which often lead to scattered knowledge, which can make it difficult to onboard new officers who may have a lot of questions. We wanted to create a simple, effective chatbot solution that can bring all this information together as a single source of truth.

What it does

OrgSage consolidates information from different organizational tools, allowing team members to access a centralized repository of knowledge. Using AI-driven insights, OrgSage can answer questions, provide resources, and help users quickly find what they need, all through an intuitive chatbot interface. Additionally, it’s customizable to the specific needs of different organizations, making it adaptable and flexible.

How we built it

We intended to built OrgSage as a SaaS solution with a strong backend that integrates with Notion, Discord, and Slack APIs. For AI, we used AI SDK and OpenAI, which is fine-tuned to process and synthesize knowledge from an organization's Notion workspace. The front end is a simple, responsive chatbot interface that requires authentication into your organization.

Challenges we ran into

Data Integration: Merging data from different platforms into a cohesive structure proved challenging, especially since each platform has unique data formats. It was very difficult to export all of the data from our Notion workspace.
Context Awareness: Careful tuning of the LLM was required to ensure that OrgSage accurately understands users' questions and provides relevant answers.
We spent a long time trying to save chat persistent across different chats with the chatbot.

Accomplishments that we're proud of

We successfully created a functional prototype that integrates with ACM UTSA's Notion workspace
Our AI is capable of handling a variety of user queries accurately, which shows our success in fine-tuning the model.
OrgSage’s customizable features allow it to serve different types of organizations, which was a key goal for us.

What we learned

We learned a lot about the challenges of integrating data across multiple platforms, as well as the nuances of fine-tuning language models for specific domains. Additionally, we gained experience in designing a scalable SaaS architecture, balancing user-friendly design with technical complexity.

What's next for OrgSage

We plan to expand OrgSage's integration capabilities to help accommodate general members of an organization rather than officers. We’re also exploring advanced AI features, like proactive suggestions based on previous questions asked in the chat conversation history, to make the tool even more useful for organizational knowledge management. Finally, we aim to release OrgSage to a broader audience and gather feedback to refine the platform further.