The Problem
A family of five is sitting in a parking lot at midnight. A social worker has 30 minutes before the family's willingness to engage disappears. Right now, that social worker is making phone calls -- to shelters that may be closed, full, or unable to serve that family's specific needs. There is no shared system for real-time shelter bed availability in most US communities.
Commercial software does not serve this space because there is no profit motive. Homelessness services operate on tight grants with no margin for per-seat licensing. The result: social workers keep personal spreadsheets, shelter coordinators answer midnight phone calls, and families wait in parking lots while the system fails them.
What This Platform Does
An outreach worker opens the app, searches for available beds filtered by the family's needs (family with children, wheelchair accessible, pets allowed), and places a hold on a bed in under a minute. The shelter coordinator sees the hold and knows someone is coming.
Without a real-time availability system, outreach workers spend hours to days making serial phone calls to shelters that may be full, closed, or unable to serve the specific client in front of them (NBC San Diego, 2025). In Los Angeles, a 2024 City Controller audit found that 1 in 4 shelter beds sat empty while approximately 5,000 people waited on waitlists -- because providers had no reliable way to know which beds were available (LA City Controller, Interim Housing Audit, 2024).
The digital workflow -- search to bed hold -- takes under 10 seconds in testing. Important: The time reduction has not yet been measured in a real-world pilot. A 90-day pilot with 5+ shelters would produce the evidence to quantify impact.
The National Alliance to End Homelessness reports that Housing First programs produce average emergency services cost savings of $31,545 per person housed over two years. While this platform does not directly house people, it is designed to reduce the time from crisis to shelter placement -- the critical first step in the housing pathway.
Who Uses It
The platform is live as a demonstration deployment at findabed.org with fictional seed data but has not yet been adopted by any CoC community for production use. It is designed to serve any US Continuum of Care, from a rural volunteer-run network to a metro area with 50+ shelters.
| Tier | Audience | Monthly Cost |
|---|---|---|
| Lite | Small CoCs, rural counties | $15-30 |
| Standard | Mid-size cities | $30-75 |
| Full | Metro areas with real-time HMIS integration | $100+ |
The platform supports emergency, transitional, overflow, DV, and reentry-eligible shelter types. v0.55 added optional support for transitional shelters serving people returning from incarceration -- a population segment where most municipal HMIS deployments do not yet model the eligibility nuances (criminal-record policy, accepting-felonies posture, VAWA-survivor protections in transitional settings). See the reentry walkthrough for the operator-facing flow.
Sustainability
- Now: Open-source under Apache 2.0. The software cannot disappear. Any deployment survives independently.
- With pilot adoption: Establish a fiscal sponsor to receive grants and contributions.
- With multiple adopters: Cost-sharing consortium where participating CoCs share maintenance costs.
- At scale: Institutional host (Code for America, a university, or a managed services model).
The zero-vendor-dependency architecture is intentional: every deployment runs on standard, widely available technology (Java, PostgreSQL, React). No proprietary component. No single point of failure.
What Funding Enables
- Community adoption support -- helping additional CoCs deploy the platform, training coordinators, supporting onboarding
- A full-time maintainer -- moving from single-maintainer open source to sustainable development with security patches and community support
- Third-party accessibility audit -- an independent audit by a qualified accessibility firm to strengthen the case for government adoption
Every dollar builds permanent infrastructure. The Apache 2.0 license means the software, its documentation, and every improvement funded by a grant remains freely available to any community that chooses to adopt it. There is no scenario where funding builds something that later becomes proprietary or inaccessible.
What Makes This Different
- Open source means permanent. Unlike vendor platforms that disappear when contracts end, this software is irrevocably free.
- Zero client PII on the DV referral path. On non-DV reentry holds, optional client name, date of birth, and notes can be collected if the CoC enables
reentryMode-- encrypted at rest with a per-tenant key, erased no later than 25 hours after the hold ends. Default-off. - DV survivor protection. No address stored, no client information, referral tokens hard-deleted within 24 hours. Designed to support VAWA and FVPSA requirements.
- Novel demand data. Captures unmet demand (searches returning zero results by population type, time, and geography). No HMIS system currently produces this data.
Recommended Grant Language
This language is sourced, specific, and defensible. It uses government-audited data (LA City Controller, 2024) and practitioner testimony (NBC San Diego, 2025) rather than internal estimates.
Funding Inquiries
Open to conversations with funders evaluating support for the project. The contact below reaches the project team directly — the team can walk through the development plan, the cost categories described in “What Funding Enables” above, and what an initial project briefing looks like. There are no committed funders or pilot partners to cite today; funding decisions and partnerships will be disclosed transparently as they land.
← Back to Finding A Bed Tonight