RentProof

RentProof is an autonomous tenant-defense agent that investigates rentals across public records and the open web, publishing cited risk packets before renters sign.

Comment 1

Inspiration

Finding an apartment should not feel like gambling with your safety, deposit, or future home. Renters often make high-stakes decisions based on polished listing pages, while the real risk signals are scattered across public housing records, city complaints, violation databases, landlord registries, advocacy resources, and the open web.

We built RentProof because renters already do this investigation manually: Googling addresses, checking city records, searching landlord names, looking for complaints, and trying to figure out whether a listing is too good to be true. RentProof turns that stressful manual process into an autonomous, evidence-driven agent.

What it does

RentProof is an autonomous NYC rental-risk agent.

A user enters a rental address and borough. RentProof investigates the property across official public data and open-web sources, then generates a cited Rental Risk Packet.

The agent checks:

  • HPD violations
  • 311 housing complaints
  • eviction records
  • multiple dwelling registrations
  • registration contact records
  • DOB safety violations
  • official NYC Open Data source pages
  • JustFix / Who Owns What context
  • tenant advocacy resources
  • open-web mentions connected to the address or landlord

RentProof then calculates a transparent risk score, explains the contributing factors, stores all observations in ClickHouse, and publishes a cited report through Senso.ai/cited.md or a local fallback page.

It is not a chatbot. It performs an investigation, records evidence, computes risk, publishes a report, and prepares a payment-gated full report endpoint.

How we built it

We built RentProof as a full-stack Next.js TypeScript application with an agentic investigation pipeline.

The architecture separates structured official data from open-web intelligence:

  • NYC Open Data SODA APIs are used directly for structured public records such as HPD violations, 311 complaints, evictions, registrations, and DOB safety violations.
  • Nimble acts as the required web intelligence layer for source verification, source discovery, open-web enrichment, official page provenance checks, tenant resource discovery, and future monitoring.
  • ClickHouse stores the event trail: investigation runs, source observations, Nimble events, risk events, published reports, and payment/unlock events.
  • Senso.ai / cited.md is used to publish grounded Rental Risk Packets when configured, with a local report fallback for demo reliability.
  • A payment-gated full report endpoint demonstrates how RentProof can monetize premium reports or future address monitoring through x402/CDP/MPP-style payment rails.

The frontend is designed to show the agent working instead of showing static results. The loading flow walks through source verification, official record checks, Nimble enrichment, ClickHouse storage, risk scoring, and cited report generation.

Challenges we ran into

One major challenge was designing the right division of responsibility between official structured APIs and open-web crawling. We did not want to misuse Nimble as a replacement for clean NYC Open Data APIs, so we made the architecture explicit: official records come from SODA APIs, while Nimble provides verification, provenance, enrichment, and broader web context.

Another challenge was making the app robust under hackathon constraints. External APIs can fail, credentials may be incomplete, and live web results can be inconsistent. We added graceful degradation so the investigation can continue even if Nimble or Senso is unavailable, while still logging failures transparently.

We also had to make the ClickHouse schema reliable. Since the app depends on event storage, missing tables caused investigation failures. We added setup and test scripts to create and verify the required tables before demoing.

Finally, we had to be careful with language. Housing-risk products can easily overclaim. RentProof uses cautious, evidence-based wording and frames its output as a public-data risk signal, not legal advice.

Accomplishments that we're proud of

We are proud that RentProof feels like a real autonomous agent, not a thin AI wrapper. It does not simply summarize user input. It performs a workflow: verifies sources, queries public records, enriches findings with the open web, stores observations, scores risk, generates a cited report, and exposes a payment-gated endpoint.

We are also proud of the sponsor integrations being central to the product:

  • Nimble is the agent’s open-web intelligence layer.
  • ClickHouse is the evidence ledger and analytics backbone.
  • Senso.ai turns structured evidence into cited, publishable reports.
  • Payment rails are represented through a full-report unlock flow.

Most importantly, the product solves a real problem. Renters, students, immigrants, and first-time movers often lack the time or knowledge to investigate a property properly. RentProof gives them a clearer picture before they apply, pay fees, or sign a lease.

What we learned

We learned that agentic products become much more convincing when the workflow is visible. The frontend should not just show a final result; it should show the agent’s decisions, source checks, evidence trail, and actions taken.

We also learned that trustworthy AI systems need strong data boundaries. Official structured records should come from official APIs. Open-web intelligence should enrich and verify, not replace authoritative sources.

Another key lesson was that citation and provenance matter as much as the model output. For a product like RentProof, the most valuable feature is not just the risk score. It is the ability to say why the score exists, where the evidence came from, when it was retrieved, and how confident the agent is.

What's next for RentProof

Next, we want RentProof to move from one-time investigations to continuous rental monitoring.

Planned extensions include:

  • scheduled re-checks for addresses and listings
  • alerts when new complaints or violations appear
  • listing price-change and delisting detection
  • deeper Who Owns What / landlord network analysis
  • neighborhood-level rental risk dashboards
  • premium PDF exports
  • agent-to-agent paid API access
  • legal aid and tenant union monitoring portals
  • stronger x402/CDP/MPP payment integration for paid reports and subscriptions

Long term, RentProof can become a tenant-defense intelligence layer: a public, cited, continuously updated system that helps renters avoid risky housing decisions before damage is done.

Built With

Share this project:

Updates