AI Research Dashboard

Problem

LevelUp Kansas is a Kansas based company who helps Kansans live their dream by finding them opportunities for associate and technical degrees. They began having problems spending too much time researching and scrambling through documents, as well as fully typing out custom follow up emails, which resulted in not being able to follow up with their clients as fast. They also didn't have an organized way to manage submissions without having multiple softwares open at once.

Solution

To solve this problem, I began asking questions and researching their process. The process begins with a client filling out a form on their website, giving detailed information about their interests, background, and goals. Following this form submission, a "navigator" (what they call their researchers), will do custom research from the internet, as well as from a folder of information that they have already obtained, to follow up with the client through email and help them choose the best career path for their needs. I then decided the best approach to solve their problems would be to create a full dashboard application that holds all of their submissions in one place, and utilizes AI to research and write custom emails for each submission. I began designing the UI for this application in Figma, before moving to the development stages. The technology I decided to use was Next.js for the framework, along with TypeScript, as I love having cleaner, type safe code. I utilized fast, modern, and advanced Next.js features such as Client and Server components for accessing and displaying data. For the database I went with a SQL based database named Supabase. For easier database manipulation and mapping, I went with Drizzle as the ORM. For authentication, I used Lucia Auth. Compared to Next Auth, or Clerk, Lucia is a lot more code to write, but I love the flexibility and the benefit of it being easily customizable and tailored to what you want. Finally for the LLM, I used OpenAI GPT-4o, also utilizing their new Assistants API.

Functionality

The form on their website is built in Sharpspring. Through Sharpspring, a postback URL is provided by Zapier that sends the form details to the dashboards custom API endpoint with a post request. This endpoint then structures the data and adds it to the Supabase database.
The dashboard is secured with authentication through Lucia, setting a session for the user giving them full access to the application. Users are stored in the Supabase database along with the users session.
The user has the ability to search through the submissions table with a search bar.
Pagination was added, limiting the amount of submissions displayed at a time. The user can switch through the pages of submissions.
In each submission, the user is presented with a fully functional headless rich text editor, along with a button to generate emails with AI based on the user submissions.
The user can mark each submission as Pending, Editing, or Approved to let other users know the status of each submission.

Future plans

Add the ability to directly send emails through the dashboard.
Add an analytics page to show realtime data of number of submissions, number of submissions to complete, or other useful information.