Ascend

Ascend's homepage
Main dashboard
Contract info highlightiing
Contract overviews and negotiation points
Outlining our workflow

Inspiration

Imagine a world where freelancers are empowered to focus on their craft, while AI takes care of the arduous tasks of analyzing contracts, understanding market trends, and negotiating optimal terms. Ascend levels the playing field, ensuring freelancers are not undervalued while fostering transparency and efficiency in client negotiations.

What it does

Ascend's key features include:

Market Rate Analysis: Our product will compare the freelancer’s requested rates against market data so the user can gauge how well they're being paid. This includes a summary section that uses Hugging Face's bart-large-cnn to pull main points from lengthy contracts the user can upload, a graph that pulls salary data for the user's occupation and location from CareerOneStop's API and compares that to what the user is being offered, and a graph that visualizes the user's current pay with their pay history. When the user uses our product, their data is stored in AWS S3.
Support Real-Time Negotiation: The AI will initiate conversations with clients and suggest reasonable terms or counteroffers through the freelancer's email. GPT-3 powers chat-based interactions and automatically drafts as a response to emails from interested clients to cut down time for negotiation, though the user can review drafts before sending the messages themselves.

Workflow Freelancer Input: The freelancer provides data about themselves including their occupation, their location, the expected rate, and any contract terms they'd like to discuss. Client Interaction: If the freelancer is in an ongoing negotiation, the AI uses GPT-3 to interface with the client by drafting emails the user can look over. It dynamically adjusts the conversation based on the client's responses, suggesting counteroffers, rate adjustments, and flexible terms. Data Visualization Dashboard: Using the data the user provided about their job, location, and contract terms, AI summarizes the proposed contract for the user to look over and we pulls historical data from the CareerOneStop API to establish baseline market rates at different percentiles in the user's state.

How we built it

We built Ascend using React JS for our frontend and Python/Flask for our backend. We also utilized Amazon S3 for our database, integrated AI using Hugging Face, GPT-3 with RAG, and Sagemaker, performed data visualization using matplotlib and data from the CareerOneStop API, and designed out application's frontend using Figma.

Challenges we ran into

Entering this hackathon, we knew that learning about the different AWS services and integrating them into our project would be a challenge, and it was. Everything took a lot of troubleshooting, researching, and plain old trying and trying again, but in the end, we were able to learn so much!

Accomplishments that we're proud of

Working with AWS for the first time. Our resilience and debugging skills. Huge emphasis on resilience and debugging skills.

What's next for Ascend

Even though this hackathon has come to an end, we're definitely not done with Ascend! We'd love to broaden our customer base and specialize our data and models not only on the freelancer population, but extend our software to business contracts other licensing agreements, as legally binding contracts are everywhere and always relevant. We would also like to improve our model to extract more specific information from contracts such as wages or potential concerns to negotiate, along with providing more in-depth market analysis using user-specific data.

In the future we would like to implement an enhanced version of our Sagemaker Market Analysis model which analyzes results from freelancing platforms like Fiverr and Upwork and predicts the most desirable pay for the contractor.