BeanBot

Inspiration

Project managers lose hours every week trying to organize tasks, follow up with developers, and generate reports. What if an agent could handle this?

What it does

The app lets employees describe their tasks to a chatbot agent that organizes them in a database. They can also update tasks through the bot. Manager accounts can do all of this as well as generate reports that describe the status of all tasks and can also trigger the agent to follow up with all employees about their tasks.

How we built it

The app was built primarily with Cursor using tailored prompts that targeted specific parts of the application.

Challenges we ran into

Learning Jac fundamentals in a few hours proved to be challenging. We decided to focus instead on building the app with JavaScript and then adapt it to Jac using generative AI.

Accomplishments that we're proud of

• Integrating agent actions to a database.
• Managing to make Cursor build an app with a language it probably has not seen much of.
• Building a functional application.
• Added MCP server through Cursor. ***

What we learned

• Vibe coding can take more time than simply writing an application the good old‑fashioned way (by hand).
• Generative models can be highly unpredictable. There's nothing like reading the documentation and using it as the main reference when building.

What's next for BeanBot

• Improving features and deploying to a dedicated server
• First round of investments

Built With

• jac