# Project export: MoveMate This document was generated by HackStack to give an AI agent context about a hackathon project. Sections are labeled with their provenance; content marked as truncated was cut to keep this document small. ## Project metadata - Hackathon: UC Berkeley AI Hackathon 2026 - Tagline: MoveMate lets students upload leases, set their housing goals, and build renter history, creating new systems of trust for fairer housing access, even without credit or a cosigner. - Devpost: https://devpost.com/software/movemate-1wck8u - GitHub: not linked - Video: https://www.youtube.com/embed/UeLOQqWXpL0?enablejsapi=1&hl=en_US&rel=0&start=&version=3&wmode=transparent - Team: contributor stats unavailable ## Devpost submission (written by the team) ### Inspiration For many college students, housing is the single largest and most stressful expense, often taking up a huge share of their monthly budget. In Berkeley, off‑campus leases move fast, and a lot of students are signing contracts for the first time with no one to translate the legal language, explain what subletting actually looks like, or walk them through the steps of renting. Students who don’t have family with renting experience, strong credit, or a cosigner are often locked out of safer options, even when they are responsible tenants. Housing is a basic need for learning, and access to it shouldn’t depend on pre‑existing credit history, wealthy guarantors, or who you know. MoveMate is our attempt to give students both guidance on the leases they sign now and a way to build a recognized renter history that makes it easier to find housing in the future. ### What it does MoveMate is a student‑focused lease companion and renter history builder. A renter uploads their lease and tells us what matters to them, and our system turns a dense contract into a clear, goal‑aware dashboard for the entire rental lifecycle, while quietly building a record of responsible renting that students can carry with them. Today, MoveMate: Extracts key terms like move‑in and move‑out dates, rent and deposit amounts, basic house rules, and whether the lease says anything about subletting, guests, or early departure. Highlights obligations and potential friction points in plain language, so students can see where the lease might conflict with their real plans (for example, wanting to sublet in the summer or leave early for an internship). Generates move‑in, during‑tenancy, and move‑out checklists to make the rental process more transparent by documenting the unit, tracking communication, and knowing what to do when it is time to leave. Tracks important deadlines with reminders for notice periods, inspection windows, and renewal dates. Provides a “Your goals for this lease” panel where students articulate what they care about (for example being able to sublet, leaving early for a co‑op, keeping their security deposit, or being able to host family/guests), and MoveMate evaluates how well the lease supports each goal with a simple fit label (good / okay / poor / unknown) and a one‑sentence explanation. On top of this, we introduce a renter rating concept: instead of another opaque credit score, MoveMate assembles an explainable, student‑centric history over time based on things like: Successfully completing leases and checklists Documenting conditions and resolving issues constructively Paying attention to notice periods and move‑out steps The goal is to prototype a new kind of trust signal that students can choose to share when applying for future housing, especially those who start college with little or no credit history or family cosigners. In the long run, this kind of renter history could give responsible students and people emerging from difficult life situations more options and give ethical landlords a way to see beyond traditional credit files. A single student can get value immediately by uploading one lease and understanding how it fits their goals and what steps to take. Over time, as more Berkeley students use MoveMate, we can also surface patterns, like how often leases are silent on subletting or impose subjective cleanliness standards, and start to build new systems of trust that make safe housing more accessible, not less. ### How we built it We centered one high‑impact interaction: a student uploads a lease, sets a few personal goals, and gets structured guidance plus concrete next steps. Our prototype pipeline: Converts uploaded leases into text and pre‑processes them client‑side to mask obvious identifiers (names, exact unit numbers, signatures) before sending anything to the backend or model. Sends the cleaned text to Claude Code with a prompt and JSON schema for extracting core fields like dates, amounts, simple rules, and any language about subletting, guests, and move‑out expectations. Combines light rule‑based parsing (for dates and amounts) with the model’s understanding of clauses to handle messy, multi‑page student leases more robustly. Makes a separate, focused model call that takes the lease text plus a short list of student‑supplied goals (for example: “I want to be able to sublet in summer” or “I might need to leave early for an internship”) and, for each goal, returns a fit label (good / okay / poor / unknown) and a one‑sentence reasoning tied to specific parts of the lease. Renders everything in a web dashboard with sections for “Key Terms”, “Your Goals for This Lease”, “Potential Friction Points”, and “Next Actions”, using a straightforward, student‑friendly visual design. MoveMate is a Vite React app with a small Node/Express API. The frontend handles the user experience, PDF text extraction, PII masking, dashboard state, checklists, lease comparison UI, and renter-history display. The backend validates requests, calls the Anthropic Messages API when an API key is configured, and falls back to deterministic demo responses when one is not. The analysis pipeline is: Lease PDF or pasted text -> browser text extraction -> client-side PII masking -> Express API -> Claude JSON analysis or demo fallback -> normalized dashboard data -> local browser storage. We separated the AI work into focused prompts: one for lease extraction, one for goal-fit evaluation, and one for lease comparison. The prompts require JSON output, plain-language paraphrasing, uncertainty flags, and no legal advice. This helped us keep the UI stable while still using AI for messy lease language. ### Challenges we ran into Messy, inconsistent lease documents. Student leases range from simple one‑page forms to long corporate contracts with scattered clauses and inconsistent formatting. Getting reliable structured output meant several rounds of prompt and schema refinement, plus a hybrid approach that uses both pattern‑matching and model‑based interpretation. Designing new trust systems responsibly. The renter rating idea is powerful but sensitive: we do not want to create another opaque score that can be weaponized against students. In this hackathon version, we focused on designing an explainable, student‑owned history concept and clearly separating it from any high‑stakes decision‑making. Any move toward landlord‑facing products would require strong guardrails, community input, and partnerships with trusted campus entities. Scoping to something we could finish. It was tempting to build everything at once, including sublet tools, roommate matching, landlord ratings, and analytics. With limited time, we scoped down to a single end‑to‑end flow that still hints at how this could grow into a broader student housing support layer and a more equitable trust system alongside existing campus services. ### Accomplishments we're proud of Shipping an end‑to‑end flow that a Berkeley student could actually use. In under 24 hours, I went from “housing is stressful” to a working prototype where a student can upload a real lease, set their goals, and walk away with concrete next steps for move‑in, mid‑lease, and move‑out. Designing goal‑aware guidance, not just another AI wrapper. Instead of stopping at extraction, I added a “Your goals for this lease” panel and a second model call that evaluates how well the lease fits each student’s situation (for example subletting for an internship or keeping a deposit), with simple labels and one‑sentence explanations. Exploring renter history as a new trust signal for students. I did not just mirror traditional credit scores. I sketched an explainable, student‑owned renter history concept that could help responsible renters with limited credit or no cosigner be seen more fairly in the future, while keeping control in their hands. Being intentional about ethics from the start. Even under time pressure, I built in client‑side redaction of obvious identifiers, clear “not legal advice” messaging, links to Berkeley support resources, and an explicit stance that any renter rating must be opt‑in, transparent, and co‑designed with the students most impacted. Keeping scope honest and impact‑driven. It was tempting to chase flashy features, but I stayed focused on one high‑impact interaction that addresses a real barrier to equitable education: understanding and surviving your first leases without a built‑in support system. ### What we learned We began with a broad question: how do we support students who are navigating housing decisions while juggling classes, work, and a cost‑of‑living crisis? In early brainstorming, we considered jobs and tuition tools, but housing kept surfacing as the place where students felt both financially vulnerable and structurally excluded, particularly those without credit history, guarantors, or family support to co‑sign and review leases. We deliberately moved away from building another listings site that assumes landlords will change their workflows. Instead, we focused on something a single Berkeley student can do today, upload the lease they already have, and on giving them clarity about how that lease fits their reality. Conversations with peers reminded us that the next lease is often just as stressful: people can be great tenants and still get screened out by traditional systems. That is what pushed us to explore the idea of a student‑centric renter history as an alternative trust signal. We iterated on prompts and data structures using a mix of real and synthetic leases, tuning how we handle vague wording and making sure explanations stay in plain language. At the same time, we mapped where MoveMate could connect to existing Berkeley resources (the Student Legal Services, the Basic Needs Center, local tenant‑rights groups) so that we are reinforcing a broader ecosystem of support rather than trying to replace it. ### What's next After the hackathon, we’d like to pilot MoveMate with Berkeley student organizations and basic‑needs staff, gather feedback on where the explanations and checklists are most helpful, and co‑design the renter history piece with students who are most impacted by current barriers (first‑gen, low‑income, international, and formerly housing‑insecure students). Longer term, our vision is a system where: Students get clear, accessible guidance on the leases they sign now They can opt in to building an explainable renter history they own and control Landlords and institutions have better ways to recognize responsible renters that don’t rely solely on traditional credit and cosigners In other words, MoveMate is a step toward new systems of trust that make it easier for students to secure stable housing and, by extension, to stay in school and complete their education. ## README (from the GitHub repository) No README available. ## Detected evidence (automated analysis) No repository was indexed for this project. Claimed technologies below could not be checked against code. - Anthropic (technology) — claimed on Devpost, not found in the code - Express (technology) — claimed on Devpost, not found in the code - HTML (language) — claimed on Devpost, not found in the code - JavaScript (language) — claimed on Devpost, not found in the code - Node.js (technology) — claimed on Devpost, not found in the code - React (technology) — claimed on Devpost, not found in the code - Tailwind CSS (technology) — claimed on Devpost, not found in the code ## Codebase structure No repository index available. ## Key source files No repository index available; no source files included.