Project Backlog

Innovative ideas with public benefit awaiting development to create positive impact through open-source technology

Submit an Idea
1

The Voice of Contracts

Problem Domain
Access to Information
Target Users
Low-literacy borrowers in Southeast Asia and Africa who sign contracts in colonial or legal languages they cannot read
Why It Matters
Prevents exploitation of vulnerable borrowers who are locked into predatory loans because they can't understand contract terms. Directly addresses information asymmetry that enables financial abuse.
Technical Solution
OCR + translation + LLM analysis + text-to-speech pipeline. Scans contract text, translates to local language (Hindi, Swahili, Bengali), uses LLM to flag predatory clauses (high interest rates, hidden fees, unfair penalties), then reads aloud the red flags in the user's dialect.
Metrics That Matter
  • Contracts scanned per month
  • Predatory clauses flagged
  • User-reported loan rejections (avoided bad deals)
  • Languages supported
  • Debt burden avoided (estimated)
2

The Harvest Guardian

Problem Domain
Access to Information
Target Users
Smallholder farmers in India and Africa without reliable internet access but with basic mobile phones
Why It Matters
Information asymmetry allows middlemen to exploit farmers by underpaying for crops. Real-time price transparency empowers farmers to negotiate fair prices and avoid predatory intermediaries.
Technical Solution
SMS-based price lookup system. Farmer sends text message with crop type and location; backend pulls current wholesale market prices and sends reply via SMS. Works offline-first with periodic data syncing for price updates.
Metrics That Matter
  • SMS queries per month
  • Number of farmers reached
  • Price improvement (farmer-reported difference between quoted price and actual market rate)
  • Geographic coverage (regions/crops supported)
3

The Real Cost of Lending Lens

Problem Domain
Financial Protection
Target Users
Low-income borrowers considering payday loans, title loans, or rent-to-own agreements
Why It Matters
Predatory lenders exploit mathematical illiteracy by emphasizing affordable monthly payments while hiding crushing total costs. Showing "you'll pay $3,000 for a $1,000 loan" prevents people from signing exploitative agreements.
Technical Solution
OCR + financial calculator. User photographs loan document; OCR extracts APR, fees, and payment terms; calculator shows total dollars paid over loan lifetime (not just monthly payment). Compares to alternatives (credit union loans, payment plans).
Metrics That Matter
  • Loan documents scanned per month
  • Average total cost revealed (vs. principal borrowed)
  • User-reported loan rejections (avoided predatory deals)
  • Estimated debt burden avoided
4

The Gig-Worker's Truth

Problem Domain
Financial Protection
Target Users
Gig economy workers (Uber, DoorDash, Instacart drivers) who don't realize their true hourly wage after expenses
Why It Matters
Gig platforms obscure true earnings by highlighting gross revenue while hiding costs. Many drivers earn below minimum wage without realizing it. Transparency empowers workers to make informed decisions about their labor.
Technical Solution
Screenshot OCR + expense calculator. Users upload weekly earnings screenshots; OCR extracts revenue data; calculator subtracts gas, vehicle depreciation, taxes, insurance; displays real hourly wage. Shows comparison to local minimum wage.
Metrics That Matter
  • Screenshots analyzed per week
  • Average true wage calculated (vs. platform-reported earnings)
  • Users who quit or reduced hours after seeing real wage
  • Platform coverage (Uber, DoorDash, etc.)
5

The Scrap Hacker

Problem Domain
Environmental Impact
Target Users
Rural residents in developing countries (India, Turkey, etc.) with access to waste materials but limited technical knowledge or tools
Why It Matters
Transforms waste into resources for communities that can't afford new materials. Empowers local innovation and reduces environmental impact in areas without formal waste management systems.
Technical Solution
Computer vision + retrieval system that identifies trash materials through camera input and retrieves engineering-grade blueprints for upcycling. Uses image recognition to classify materials (bottles, tires, metal scraps) and matches them to a curated database of DIY construction/upcycling instructions.
Metrics That Matter
  • Blueprints retrieved per month
  • Materials successfully repurposed (user-reported)
  • Geographic spread (number of villages/regions using tool)
  • Cost savings for users (estimated value of items built vs. buying new)
6

The Green Grant Navigator

Problem Domain
Environmental Impact
Target Users
Small nonprofits and community groups working on climate resilience projects who can't afford grant writers
Why It Matters
Climate funding disproportionately goes to large organizations with grant-writing capacity. Small community groups doing frontline climate work can't access capital. Leveling the playing field gets resources to those closest to the problems.
Technical Solution
RAG system trained on federal climate funding databases (EPA, FEMA, DOE). User inputs project description; system searches for matching grants, explains eligibility requirements, and drafts narrative sections of grant application using templates and best practices.
Metrics That Matter
  • Grant searches per month
  • Grant applications submitted (user-reported)
  • Funding secured (user-reported)
  • Geographic/demographic diversity of users
  • Total climate funding redirected to small orgs
7

The Scope Creep Shield

Problem Domain
Access to Justice
Target Users
Freelancers and gig workers (designers, writers, developers) who face scope creep and payment disputes
Why It Matters
Most freelancers work without proper contracts or sign client-biased agreements because they lack legal knowledge. This leads to unpaid work, scope creep, and payment disputes. Levels the playing field.
Technical Solution
LLM-powered contract analysis that reviews freelance agreements, flags problematic clauses (vague scope, no kill fee, missing IP rights, unlimited revisions), and suggests protective language. Acts as "bad cop" advocate for freelancer interests.
Metrics That Matter
  • Contracts reviewed per month
  • Red flags identified per contract
  • Protective clauses added (user-reported)
  • Payment disputes avoided (survey data)
8

The Eviction Shield

Problem Domain
Access to Justice
Target Users
Low-income tenants facing eviction or unsafe housing conditions
Why It Matters
Tenants lose cases not because they lack defenses but because they don't know their rights or how to assert them. Landlords send legally dubious threats knowing tenants won't fight back. Empowering tenants with legal knowledge prevents wrongful evictions.
Technical Solution
Location-aware RAG chatbot pulling from local tenant protection laws. User enters ZIP code and issue; system explains specific rights under local law, provides step-by-step guidance, and generates legally-sound demand letters to landlords. MVP covers 5 cities with strong tenant protections.
Metrics That Matter
  • Chatbot queries per month
  • Demand letters generated
  • User-reported evictions prevented
  • Habitability improvements achieved
  • Cities/jurisdictions covered
9

The Disaster Prep Agent

Problem Domain
Community Resilience
Target Users
Households in disaster-prone areas (hurricanes, earthquakes, wildfires) who need actionable emergency plans
Why It Matters
Information paralysis kills people during disasters. Dense government documents aren't actionable in emergencies. A simple, hyper-local plan increases survival rates and reduces panic.
Technical Solution
RAG (Retrieval-Augmented Generation) system pulling from FEMA guides and local emergency management data. User enters ZIP code and disaster type; system generates customized, printable one-page emergency plan with evacuation routes, supply checklists, and emergency contacts. Offline-cacheable.
Metrics That Matter
  • Plans generated per month
  • ZIP codes covered
  • Downloads during declared emergencies
  • User-reported preparedness improvements (survey data)