TenderYes | Devpost

TenderYes can accept text or pdf input.
TenderYes can easily detect if a tender is present based on the input.
TenderYes produces three email drafts with varying levels of formality, allowing users to select the draft that works best for them.
TenderYes searches the internet to find the best, most relevant cases and laws.

Legal teams don’t struggle because they lack tools. They struggle because every day starts with a flood of unstructured information. Emails, PDFs, and research requests pile up, and simply figuring out what to do next takes too long. We wanted to explore how AI could act less like a chatbot and more like a reliable case team member that helps lawyers move faster without losing control.

TenderYes is an AI Orchestrator designed to turn messy legal inputs into clear, actionable outputs.

Users can upload a PDF or paste in text, and TenderYes will first tell them if a tender has been detected, then routes the task to the appropriate AI specialist. Today, it includes:

Client Communication Guru: which generates three client-ready emails at different levels of formality.
Legal Research Pro: which uses grounded AI research to quickly surface relevant case law.

Each specialist returns focused, decision-ready results, helping legal teams respond faster and research smarter.

Users can swipe right or left to select the results of TenderYes that will work best for them!

We built the app as a mobile-first workflow that turns a tender offer email (in pdf or txt format) into fast, actionable decisions. A React Native (Expo) frontend lets users paste or attach an insurer’s email and immediately see swipe-style cards. Behind the scenes, a lightweight backend securely calls the Gemini API in three steps: first to extract and summarize the tender details (amount, deadline, conditions, and supporting evidence); second to generate three client-ready email drafts at different levels of formality, allowing users to select their most preferred draft; and third to search the internet for the most relevant resources for the user using Gemini's grounding feature. This separation of UI, orchestration logic, and AI calls made the system fast to build, easy to demo, and realistic for a high-volume legal workflow.

We ran into a lot of challenges using the Gemini API. AI isn't perfect, so we had to precise write our prompts to prevent it from hallucinating or not returning the right answer consistently. Additionally, we were very limited with the amount of calls we could make with the free tier of the Gemini API. However, we were determined to power our project with the Gemini API, so we had to get creative with our solutions. Our result was very innovative and impressive considering the short time frame.

Accomplishments we're proud of:

Getting more experience with the Gemini API.
Designing and implementing a full-stack project as a team in a short amount of time.
Successfully routing unstructured PDFs and text to the correct AI specialist
Generating client communications that balance professionalism and empathy
Producing fast, grounded legal research that would normally take hours
Building a foundation that can scale into a full AI case team

What we learned: We learned how we can utilize AI in everyday life to make other people's lives easier. We also learned a lot about coding a project that produces AI for a result. We had to figure out how to keep the factors controlled when we could to account for the sometimes unpredictable nature of AI.

What's Next?: We plan on perfecting the prompts, being able to email directly from TenderYes, and add more features, like a Voice Bot Scheduler, or a Evidence Analyzer. We also want users to be able to have all of the info that they "swiped right" on in a single pdf that is available for download. We also want prompts to be able to be adjusted based on user input to give users the most customizable and personal experience.