Civic Concierge for San Francisco

Civic matching, starting from what the household already qualifies for. With the household's permission, the system records the benefits a household is already eligible to receive, then connects that household only to programs those benefits already cover.

1. 1.Footprint: with permission, gather public lines already attached to the household (CalFresh, WIC, LIHEAP, Medi-Cal, local rental and mortgage aid, veterans' benefits) into one annual baseline. Tag who they are: families, elders, veterans, children, disabled residents.
2. 2.Qualifying match. Need + footprint + local inventory. Only programs the person qualifies for, in this city. First-time buyer and down-payment help, rental and mortgage aid, rapid rehousing, shelter beds, food banks, weatherization, coverage navigation, training after base needs.
3. 3.Pathways with a defined next step, rather than a standalone referral. Housing is sequenced from crisis response to stabilization, assistance, and ownership. Nutrition, energy, health, transportation, and employment follow the same structure. Each completed step updates the household record.
4. 4.Ledger. The system records dollars expended, people served, community category, and district, not referral counts. The purpose is to show what public funds purchased.
5. 5.Civic Brain Trusts. Expert groups organized on the nine EcoQuest pillars, applied here to social-service delivery. They set cost and outcome targets used by the Bounty Board.
6. 6.Bounty Board. When cost-per-unit, cost-per-calorie, cost-per-kilowatt, or time-to-housing misses a published target, qualified solvers may submit a method. The relevant Brain Trust reviews feasibility. Accepted playbooks remain in a jurisdictional repository. San Francisco can reuse them. Other departments and, later, other cities can license them so the same miss is not funded twice. Money from licensing those playbooks, if it appears, returns to the city's giving program (the next navigator cohort, clinic slot, or prevention rail). Any equity interest is held in a separate company. The board itself is software.

The below-market instrument would finance production inside San Francisco through separate companies and charitable affiliates. It would not capitalize the application, purchase tokens, or create a claim on any household's public benefits.

• Housing production and retrofit aimed at lowering unit cost.
• Community-scale food production and wholesale contracts CalFresh households may join.
• Weatherization and local power that cut the energy line on the footprint.

A 30 percent reduction below market cost is a planning target drawn from comparable licensed production elsewhere, not a guarantee and not a result recorded in this city. If later verified, $1 of public expenditure would model as approximately $1.43 of goods, and a $298 one-person SNAP allotment would model as $426 of food. Those figures are planning analogs, not measured San Francisco results. The measured ratio would be published after operations begin in this city.

The intended sequence is stabilize → participate → produce → own: stabilize (housing, food, and energy), participate (employment), produce (community production), and own (home and energy interests held in separate legal entities). Employment services follow stabilization because job placement is more durable when those base conditions are in place.

The software layer above is not a stripped civic copy. The deployment carries the civic tools Volume I specifies for the platform, each aimed at a gap this city has already published: a same-day need that cannot wait for a callback, surplus that is discarded at close while a senior waits, and operators with no lawful way to see committed demand.

The board runs at the speed of today, beside the regular programs (with their applications, windows, and waitlists) already described. A shelter bed for tonight, a meal before 6 p.m., a ride to a licensed appointment: those needs do not survive a phone tree, and this city's surplus kitchens and opted-in restaurants should not have to discard what a posted line would have claimed. Zones are neighborhood bands large enough to route a window and small enough to be useful, never an address.

Every tool above runs inside the boundaries this filing already states, and the boundaries do not loosen for speed.

• The Daily Needs Board does not determine statutory eligibility, disburse public funds, or replace 211, HMIS, or a provider's system of record. It can interface where authorized, the same sequencing as every other integration in this filing.
• The software does not cook the meal, drive the ride, or hold a bed. Food-code compliance, licensure, insurance, and title stay with the operator who accepted the ticket.
• No digital token has a claim on any meal, ride, seat, or donated dollar, and Network Token proceeds are never mixed with ticket fulfillment or gift dollars.
• Badges for operators are revocable proofs of standing. They do not rank a household on any public board and do not substitute for a license, a bond, or an agency's procurement rules.
• A household above a benefit line and a household on a benefit line can both post. Immediate need is not a means-tested status, and the civic result set may still appear beside the ticket.

What the daily tempo adds is a measured object this city does not currently publish: the unmatched rate, by need, by community tag, by zone, by day. That number, not an anecdote, is what rolls to the Bounty Board, and closing it is a metric a Brain Trust can set and a solver can be paid against.

The map sits on the same authorized footprint already described in this monograph. A resident states a need. The software records jurisdiction, qualifying programs, and the next step. Over time those records, stripped of names and addresses in any public or partner view, become a living picture of demand by neighborhood, by pillar, and by population tag: families, elders, veterans, people with disabilities, children.

What a city official, a Brain Trust, or a licensed provider would see is not a person. They would see blocks: how many qualifying households in this district need a nutrition package this month; how many need a same-week mental-health intake; how many need a prevention visit before an eviction filing; how many need a child-care slot tied to a work schedule. The unit of commerce is the block, not the file.

The map is a coordination surface, not a surveillance product. It does not determine eligibility. It does not disburse benefits. It does not publish a household. It shows, in the aggregate, where this city's existing programs and this city's existing businesses can meet each other.

Three audiences use the same picture for different work.

Households use it only as a next step: a clinic appointment, a food pickup, a housing rail, a workforce seat. They do not see other people's need.

The City and County, Human Services Agency, Department of Homelessness and Supportive Housing, and nonprofit partners use it to see load: which districts are absorbing more demand than inventory, which rails are stalled, which population tags are underserved relative to the dollars already assigned.

Local providers and small businesses use it to see work they are licensed to do. A food cart or community kitchen that already holds a health permit can see a nutrition block. A holistic or community clinic that already holds the right license and insurance can see a care block. A bonded contractor can see a repair or weatherization block. A child-care operator can see a slot block tied to working families. The invitation is not "become a social-service agency." The invitation is "bring the service you already know how to run, under the rules you already live under, to people this city has already found eligible."

That is the empowerment and symbiosis this section aims at. The city does not invent a new vendor class. It opens a visible, lawful market in which the people who already feed, house, treat, and employ San Franciscans can take a share of organized demand instead of waiting for a separate RFP that never names them.

Where households opt in, and where program rules allow, qualifying demand can be assembled into a service block: a defined unit of work, a defined period, a defined neighborhood or population, and a defined public or pooled source of payment. Licensed providers bid on the block.

A bid is an offer to deliver the specified service at a stated unit cost, on a stated schedule, under stated standards. The software records the offer. The relevant Brain Trust or administering agency reviews fitness: license in good standing, bond or insurance as required for that class of work, complaint history, capacity, and whether the offer meets the metric on the block. Award follows existing public-procurement and program rules. The Civic layer does not replace those rules. It makes the work visible enough that a small provider can find it.

Examples of blocks this city could stand up without inventing new benefit programs:

Nutrition. Qualifying CalFresh and related households in a district who opt into a pooled purchase. A licensed kitchen, cart, or grocer bids to deliver a weekly package whose retail equivalent meets or exceeds the allotment the household already receives. The household keeps choice. The provider gets predictable volume. The city can see meals delivered against the footprint instead of only issuances at a retailer.

Care. Qualifying residents who need a first appointment, a follow-up, or a community-based session. A licensed clinic bids on a block of intakes for a month. Payment follows completed visits, not marketing. Standards of practice remain those of the license.

Housing-adjacent work. Prevention visits, habitability repairs, weatherization, or move-in kits attached to households already on a housing rail. A bonded contractor or community group bids on a defined package, not an open-ended retainer.

Child care and work. Slots timed to the hours a parent must keep a job. A licensed provider bids on a block of seats rather than hoping walk-ins appear.

Each block is a contract the community can see close: how many units were offered, how many were delivered, to which population tags, at what unit cost, against what public or pooled dollar. That is the ledger already named in this monograph, applied to local commerce instead of only to large agencies.

Autonomy for a small provider is not a waiver. It is the right to bid on work the provider is already allowed to do.

Every provider on the map holds the license, permit, bond, and insurance that class of work already requires in California and in San Francisco. A food vendor meets health code. A clinic meets professional and facility rules. A contractor meets state licensing and insurance rules. A child-care operator meets Community Care Licensing. The software does not create a parallel license. It records the existing one and refuses a bid that cannot show it.

Standards are ordinary and public: scope of work, complaint path, non-discrimination, data-use limits, and the right of the administering agency to pull a provider that fails them. Consistency is the point. A household should receive a service that looks like a service, not a favor. A provider should know the bar before they bid.

What the platform adds is not a new regulator. It adds a lawful way for a licensed neighborhood business to find and bid on that work.

Public benefits arrive as thin, separate lines: an allotment, a voucher, a utility pledge, a one-time prevention check. Each line, used alone at retail, buys less than the same line used with others who have the same need and the same qualifying status.

Where law and household consent allow, the Civic layer can group those lines into a package: a nutrition package, a prevention package, a first-month housing package, a care package. The household still holds the underlying benefit. The software accounts for the share. The provider delivers against the package, not against nine separate invoices.

This is the same pooling logic already stated for wholesale food purchase and for rent or build contracts. The map makes it usable by the businesses that would actually cook, treat, repair, and staff the package. Innovative use here does not mean a new public program. It means using an existing allocation as a committed volume so a local provider can price below walk-up retail and still operate.

If a later, separately financed production method lowers the unit cost of food, energy, or housing, the package is how that saving reaches the household and how the surplus is measured. A planning target of stretching an allotment is not a guarantee. It is a metric the block can be written to test.

The Civic map is a board against guessed demand. Qualifying need, once opted in and stripped of household identity in any public view, becomes a posted block. A licensed provider does not invent a market. They read one. Production follows a bid, not a hope.

What "save a business" means here is not a subsidy for an empty storefront. It means a shop that already holds a license can see enough committed volume to keep the crew, the lease, and the insurance. A food cart that already meets health code can bid on a nutrition block instead of waiting for lunch traffic that rain or a closed office tower can erase. A community clinic can staff a month of intakes because the block says how many first appointments this district will pay for. A bonded contractor can order material after award, not before.

Predictable demand is what saves the shop. The public dollar is already in the city. The household is already eligible. What the shop lacked was a way to see those two facts as work. The board is how those two facts become work.

What "encourage the next business" means here is a change in the order of the bet. A person who would open a kitchen, a repair shop, or a licensed care room in this city now faces rent, build-out, payroll, and no picture of whether the neighborhood will use the service. Many do not take the bet. Those who do often place the shop where rent is cheap rather than where the unmet block is largest.

A posted board lets the founder see which blocks this district cannot fill with current vendors. They can size the shop to a real unit of demand. They can write a business plan against a metric the city already funds (meals, intakes, prevention visits, child-care seats) instead of against a slogan. Usefulness is no longer a side effect of surviving the first year. It is the reason the shop exists.

The software does not pick winners. It does not waive license, bond, or standard. It does not promise volume to a shop that cannot meet the bar. It publishes the need in a form a serious operator can underwrite.

Blind production is cooking, staffing, or building before the city or the household has said yes. It wastes food, wages, and material. It also wastes public patience: residents see empty programs and full dumpsters and conclude that "services" do not work.

On this board, production starts after three conditions are true. The household has opted into the package or the step. The administering agency or Brain Trust has accepted that the block is lawful and that the bidder is licensed and bonded for that class of work. The award is recorded.

Only then does the kitchen cook, the clinic open the hour, the contractor buy the lumber. The ledger then records what was delivered against what was paid. A shop that cannot close the block does not get the next one. A shop that closes it at a better unit cost becomes the proof that this neighborhood can do the work.

The reward is not a speech. It is the next block, the retained crew, and, where the City sets the rule, a recorded remainder that can recycle into the following cycle or into a community stake in a local operating entity. Impact is the metric the bid was written to: completed steps, population tags served, unit cost against the public dollar already assigned.

A provider that delivers more meal value than a retail allotment, more kept appointments than a walk-in clinic, or more prevention visits before a lockout, shows that on the ledger. That record is what other stakeholders read when they decide to bid, to hire, or to open a second location. Community impact is therefore not a separate page. It is how the shop stays in business.

Fewer dark storefronts on corridors that already have qualifying need. A reason for a licensed worker to open here instead of elsewhere. A vendor bench wider than the at least 232 providers Crankstart mapped in 2024, without a new class of unlicensed vendor. A public picture that says: this is the work, these are the shops that bid, this is what closed, this is what remains.

Three returns are in view. Service return: more completed steps per public dollar already in the city. Capacity return: neighborhood businesses that can hire and stay open because demand is visible and contracted instead of episodic. Civic return: a map the City can show residents and funders (need, bids, what closed, what remains) so merchants, clinics, houses of worship, block groups, and employers step into a shared condition instead of a private emergency.

A remainder that appears when a block closes under cost does not become a private claim on public funds. It is recorded on the ledger and, under rules the City would set, can be recycled into the next block, into a community stake in a local operating entity, or into the giving mechanism already described for playbooks and licensed methods that this city can keep.

Retail groceries are not the only lawful way a nutrition allotment can feed someone. Once a household is on the Civic footprint, and once a nutrition rail exists that records eligibility, consent, and delivery, this city can ask a different question: can a meal that is already being cooked in a licensed kitchen be the unit the allotment buys?

The partners for that question are not new city departments. They are producers, existing kitchens, and, where the brand and the license allow, franchise locations that already run a line, already pass health inspection, and already know how many meals they can prepare in an hour. A partnership here is a contract to use spare or scheduled capacity against a posted nutrition block, not a request that the shop become a social-service agency.

Three funding sources can pay for one meal without mixing their legal character.

The household's qualifying allotment (CalFresh or a related nutrition line) remains the household's. The software accounts for the share. It does not seize the benefit.

A local economic contribution (a merchant association, a district, an employer, a philanthropic match already in this city's pattern) can stand beside that allotment to buy down the unit cost of the meal. That contribution is private or civic capital. It is not a claim on the public benefit.

The kitchen's own production (volume from a posted block, use of a line that would otherwise sit idle between rushes, or a franchise's existing commissary) is what makes the meal cheaper than a walk-up retail grocery basket of the same calories. The kitchen is paid for meals delivered, not for sitting in a meeting.

Together those three can subsidize production without pretending the City has invented a new entitlement. The entitlement is the one the household already holds. The partnership is how that entitlement meets a kitchen that already exists.

A hot meal is not claimed here as a present CalFresh rule in every case. Federal and state nutrition rules decide what an allotment may buy. This monograph does not rewrite those rules. It builds the pathway first: identity of the qualifying household, consent, a licensed kitchen, a posted block, a ledger of meals delivered against an authorized dollar.

Once that pathway exists, two things become possible that are not possible while the city only sees issuances at a retailer.

Fiscal. The city can test whether a prepared meal, bought at kitchen cost plus a posted margin, delivers more nutrition per public dollar than the same dollar at a corner store. If it does, the case for allowing or expanding hot-meal use of an allotment is a measured case, not a slogan.

Social. A meal that is hot, local, and regular is a different service than a bag of goods the household must still cook with time and equipment they may not have. Elders, workers on split shifts, and families in rooms without a stove are the households for whom that difference is the program.

Compliance. Eligibility, licensing, food safety, and accounting stay in their present lanes. The kitchen keeps its health permit. The household keeps its rights-holder status. The agency keeps eligibility. The software records the match and the delivery. A future rule change, waiver, or restaurant-meals authority, if the State or the federal program later grants it, would attach to a pathway that already works, instead of to a pilot that has to invent intake, vendors, and a ledger at the same time.

A franchise or a neighborhood kitchen will not tool a hot-meal line on a rumor of need. They will tool it if a block says how many qualifying meals this district will pay for this month, under a license they already hold. That is the same predictable-demand logic as the rest of this section, applied to meals.

Upon advancement, a first kitchen partnership would be designed with the City as a measured block: licensed facility, published food-safety standard, household opt-in, ledger of meals against the footprint. Whether that block uses an existing allotment, a local match, or both would be stated under diligence. Operators would be unveiled then. No vendor is named in this public filing. No change to federal nutrition law is asserted as already won.

Boundless asks whether an approach will shift conditions for San Franciscans, not only deliver another service. A visible map of need, with licensed local providers bidding on blocks that households already qualify for, is that shift in commercial form. A predictable board for need is that shift for commerce. It does not replace CalFresh, Coordinated Entry, DAHLIA, or the Department of Public Health. It does not replace the market. It stops asking the market to guess. It does not name vendors in this public filing. It does not waive license or bond.

It gives this city a way to turn fragmented demand into work a neighborhood can do, under the rules a neighborhood already knows, with a ledger that shows whether the dollar bought more than it used to, and whether the extra stayed here.

Upon advancement, the first blocks proposed for design with the City would be named under diligence: one nutrition block and one care block, sized small enough that a neighborhood operator can bid without becoming a city department, each with a published standard, a bidding rule, and a public metric. Partners and operators would be unveiled then. This section is the logic those blocks would follow.

San Francisco does not lack programs. It lacks a coordination layer that treats public expenditure as seed capital and the resident as a person moving along a defined pathway.

Four structures would change. Opportunity: a household would see the programs for which it qualifies in this county, with the next step attached, reducing the advantage of applicants who already know the system. Norms: benefits would be described as time-limited support toward self-sufficiency. Systems: agencies would share a ledger of dollars, people by category, and delivered value. Unit economics: production methods, held outside the civic software, would be aimed at the cost of housing, food, and energy, the inputs that currently require a new appropriation each year. Methods and vendors would be named only if an Award is made.

Direct service (navigation, matching, pooling) is the foundation. The ambition is a city that can stabilize without rewriting the same emergency. Software funds, if any, would size to the coordination layer the City does not have. Production capital, if later used, would sit in separate entities beside existing public flows (the $441.6 million voucher book, about 111,700 CalFresh residents, the 20,341-unit shortfall, and the 7,973 people counted on one night), not as a request stated in this draft.

The City's May 2024 release put sheltered people at 48 percent of the 2024 count, up from 36 percent in 2019; HSH's May 2026 preliminary report put the share at 57 percent. Home by the Bay (the City's 2023–2028 plan) and Breaking the Cycle (a 2025 public-private fund, about $37.5 million at launch) are context the layer would sit beside, not evidence it works. The gap they name is inflow and exits, not the absence of outreach. Housing Authority voucher money is almost fully used: 99.8 percent budget utilization per HUD's dashboard, cited by the Budget and Legislative Analyst in June 2026; a spending figure that includes project-based vouchers, not proof a household can find a unit; the failure is unit match and time. Human Services already runs CalFresh and Medi-Cal. We do not replace those systems. Interfaces would open where authorized. 211 already exists. It stops at a number. The specification is that the layer would keep the footprint current after the call.

Intended public counterparts after an Award, not current signatories: HSH, SFHA, SFHSA, MOHCD, and 211 for authorized interfaces; licensed clinics and community groups as paid subgrantees if they opt in; production affiliates in separate entities; an independent evaluator; a public dashboard. This submission does not attach letters, MOUs, or vendor names.

We will not determine eligibility, disburse SNAP or Section 8, replace HMIS, hold title to housing or food inside the software or any token, mix token proceeds with benefit dollars, or count Bay Area work above the 25% cap. Civic matching is San Francisco.

The Civic layer extends methods this city and this field already use. Coordinated entry already reduces repeated crisis use; this proposal adds cross-program qualifying match, pathway steps, and a value ledger. Unused benefits remain large when navigation is fragmented. Production-cost reduction can increase the goods a transfer buys in food, weatherization, and construction; those effects will be measured at San Francisco sites. Employment attached after housing is more durable than employment attached from unsheltered settings. Public ledgers change allocation when departments use them. Upon advancement, a twelve-month district pilot will convert this logic into city-measured results against published HSH and SFHSA baselines.

Year-one numbers we will publish: footprints opened; match-to-enrollment on housing and CalFresh; time from stated need to first completed step; shelter exits among footprint households versus the city baseline; delivered value per public dollar, even when that number is $1.00.

Financial sustainability. Production assets, if formed, amortize in separate San Francisco entities. Operating surplus, if any, remains in those entities. A defined share of surplus and of license fees from reused methods is specified to replenish the city's giving mechanism. After year five, software operations are intended to run on a City contract and normal operating income rather than a recurring philanthropic grant of the same size.

Institutional adoption. If HSH, the Housing Authority, and the Human Services Agency adopt the household record and ledger, the layer continues after the award period. That adoption is a year-four success condition. The City then owns the picture of who was served, by category and district, and whether exits rose.

Community ownership and three yields. Title, if issued, sits in housing, food, or energy entities the household or neighborhood can hold, not in the application. The return to San Francisco comes in three reusable forms: utilization (programs already funded are completed instead of abandoned halfway); method (a solved miss becomes a procedure other programs can reuse); and asset (where production surplus exists, a defined share funds the next prevention rail). Social services remain a stepping stone. The public program does not disappear. The household can step off it because a wage, a unit, or a share in a local entity exists.

Community ownership. Households that reach ownership would do so through separate legal entities. Playbooks would remain reusable in this city.

Norms. Public benefits are framed as time-limited support toward self-sufficiency, not as a stigma.

After year ten the test is whether the city is still creating discrete new programs each budget cycle, or whether lower unit costs and a maintained household record have made existing expenditure sufficient.

Homelessness in San Francisco is not distributed evenly, and neither is access to help. Black residents are a small share of the city's population but a far larger share of the people counted on census night, a snapshot ratio from the Point-in-Time demographic profile, not a claim about who becomes homeless or why. Families with children were the fastest-growing group in the 2024 count (+94 percent over two years, 405 families on census night, city figures). Older adults, disabled residents, and veterans carry the oldest average stays in the system. Language tells the rest: tens of thousands of San Franciscans read little or no English, and a navigation tool that only works in English and only works on a smartphone excludes exactly the households most likely to be missed.

The design answers with what it does, not with who it praises. Access is deliberately multi-channel: the footprint and match work on a plain phone, through a navigator, and on paper the agency already holds, so eligibility does not depend on owning a device or speaking English. The Daily Needs Board is a public count by category, tag, and zone, so a district where meals or beds go unfilled is visible, not hidden inside an average. The ledger tags dollars and completions by community category and district, so the funder and the City can see whether the people the count under-serves are the people the money reaches. None of these fields diagnose a household. All of them make disparity measurable, which is the precondition for correcting it.

Equity also runs to who gets the work. Service blocks are sized for licensed small shops (a kitchen, a contractor, a family-run operator), not only for the largest agencies, and the Bounty Board is where an unfilled block in an underserved district stays visible until someone takes it. The pathway sequence (stabilize, participate, produce, own) is an equity instrument as well: it is aimed at the documented gap between receiving aid and holding an asset, the gap that keeps subsidy permanent for households that could own instead.

Stated plainly, because a reviewer will read for it: this filing does not claim that a software layer removes structural inequity. It claims that a public ledger, a visible map of unmet need by district and community category, and a pathway that ends in ownership give the City and its funders the instruments to see and act on inequity they currently cannot measure in one place.

Systems. Public expenditure in San Francisco is large and siloed: HSH $832.7 million in FY2026–27; SFHA $441.6 million in voucher authority; approximately $225 million in CalFresh benefits. Coordinated Entry ($5.9 million in FY2025–26; on the record, 10,314 assessments in the year ending March 2023, and a January 2026 account citing HSH reported more than 19,000 people active in it in the prior year, about 9 percent of them moving into permanent supportive housing) matches people to homeless-housing opportunities only. An assessment is a scored questionnaire used to prioritize a queue, not a housing count; the supported gap is assessments without placements. This proposal adds a cross-program allocation ledger and a qualifying match that extends beyond that door to first-time homebuyer and down-payment programs, rental and mortgage assistance, CalFresh, LIHEAP, coverage navigation, and employment after base conditions are in place.

Policies. A public ledger tagged by community category (families, elders, veterans, children, disabled residents) and district gives the Board of Supervisors, Our City, Our Home, and department heads a basis for allocation other than last year's formula.

Norms. Benefits are framed as bridge capital toward contribution and ownership, which is a change in the civic story the city tells about aid.

Opportunity structures. Information advantage shrinks: a household sees the programs for which it qualifies in this county, with the next step attached. Direct service (navigation, matching, authorized data assembly) is not the end state. It is the foundation required for the structural change.

Proportion. Software funds, if awarded, would size to the coordination layer the City does not operate today. Production capital, if later sized, would sit in separate entities that can change unit cost in this market. Combined, they would sit next to existing annual flows measured in the hundreds of millions, not in place of them. National and international replication is out of scope and is stated as such.

City context used in this plan. The City's May 2024 release put sheltered people at 48 percent of the 2024 count (36 percent in 2019); HSH's May 2026 preliminary report put the share at 57 percent. Home by the Bay (2023–2028) and Breaking the Cycle (a 2025 public-private fund, about $37.5 million at launch) are context the layer would sit beside, not evidence it works. The 2025 Needs Assessment explains persistent inflow (approximately three entries per exit). The 2026 Civil Grand Jury explains the data and oversight gap beside $487 million in nonprofit contracts. SFHA's 99.8 percent budget utilization (HUD dashboard via the Budget and Legislative Analyst, June 2026; includes project-based vouchers) beside a 20,341-unit affordable shortfall explains why a voucher is not the same as a home. The Food Security Task Force explains why only 39 percent of food-insecure residents under 200 percent of FPL receive CalFresh. DAHLIA, 211, Coordinated Entry, and SFHSA are treated as systems to interface with, not systems to replace.

Communities served, in order. People in the homelessness response system (7,973 on PIT night; 23,561 across FY2024–25); families (the PIT cohort that increased); veterans; elders, disabled residents, and children on the same footprints; CalFresh and voucher households not yet in crisis, to reduce the 9,552 first-time entries.

Execution sequence. District pilot (months 0–9) → citywide civic rails (months 9–18) → first loan-funded production sites inside the city (months 12–36) → departmental adoption of the ledger (months 24–60). Match-quality gates control expansion.

Path to San Francisco presence. Upon advancement: a year-one office in the named pilot district; a San Francisco managing director; navigators hired from organizations already serving that district and paid as subgrantees; interfaces opened with HSH, SFHA, SFHSA, and MOHCD as those departments agree; an independent evaluator retained in the city; and letters of collaboration executed against the live scope. Presence is a commitment of advancement, not a claim that those documents are attached to this public monograph.

Stage. This municipal deployment is early. The application therefore relies on analogous evidence and a testable logic, which the rubric expressly allows.

Logic. If households cannot see the programs they already qualify for, utilization and exits lag expenditure. If unit costs of housing, food, and energy do not fall, each year's appropriation buys the same volume. If work is attached before housing, food, and power are stable, placements do not hold. If impact is counted as referrals, the city cannot tell what taxpayers purchased.

Analogous evidence, with limits. Coordinated entry is already required of Continuums of Care; San Francisco operates it; this proposal extends its logic across benefits rather than inventing a front door. CASPEH (UCSF BHHI, 2023) found median pre-homeless income of $960 per month and named housing cost and income loss, not a preference for unsheltered life. SNAP access research and San Francisco's own 39 percent CalFresh take-up among food-insecure residents under 200 percent of FPL show benefits left unused. Furman / Strochak voucher research shows national lease-up of 31 percent at 60 days and 61 percent at 180 days; time and units, not authorized dollars, bind tight markets. Analogous operating evidence: comparable U.S. production projects have reported savings of up to 30 percent versus conventional cost. That figure is the basis for the planning target in this proposal's model. It is not reported here as a San Francisco result. Weatherization and community-production literature likewise show that public dollars can buy more physical output when unit cost falls; those percentages will be verified locally.

What will count as results here. Year-one public metrics: authorized footprints opened; match-to-enrollment on housing and CalFresh rails; time from stated need to first completed step; shelter exits among footprint households versus the city baseline; delivered-value ratio versus one public dollar, published even when the ratio is 1.00. Production-cost reduction is reported only after measurement at San Francisco sites.

Financial sustainability. Software funds build the civic instance. Production capital, if used, capitalizes separate entities that can amortize. License fees from reused bounty methods, and a defined share of production surplus, replenish prevention, clinics, and ownership pathways. After year five, software operations are intended to sit on City contract plus network operations, not on a recurring philanthropic appropriation of the same size. The Network Token has no claim on SNAP, Section 8, or local aid.

Institutional adoption. Year-four success condition: HSH, SFHA, and SFHSA can run reports from the ledger. If they cannot, durability has not been achieved, regardless of software quality.

Policy change. Category-tagged public reporting is the mechanism for allocation reform at the Board and Our City, Our Home.

Community ownership. Households that reach ownership do so through separate legal entities. Bounty-Board playbooks remain reusable in this city.

Five-year horizon. Months 0–9 pilot; 9–18 citywide rails; 12–36 production; 24–60 adoption and first loan repayment if surplus exists. Beyond year five: City-operated coordination layer, amortized assets, licensed playbooks. Beyond year ten: the test is whether the city is still creating niche annual programs or whether existing spend plus lower unit costs is sufficient.

1. 1.Production methods for food, construction or retrofit, and energy would be licensed into separate operating entities if an Award is made. They would not live inside the civic software, any digital token, or any public-benefit disbursement. Vendor identities are not part of this public filing.
2. 2.Cost reduction is a planning target, not a San Francisco result. Comparable U.S. production projects have reported savings of up to 30 percent versus conventional cost. Those figures are analogous only. They will be treated as local results only after measurement at San Francisco sites. Supporting documentation, if requested after an Award decision, would be provided in confidence.
3. 3.A software layer that scales and automates assignment. Methods that already work on individual sites do not, by themselves, close a citywide homelessness, hunger, or energy gap. Scale requires matching households that qualify in this county to the output of those methods, pooling small public allotments where lawful, publishing cost-per-unit and cost-per-calorie, and routing missed targets to the Bounty Board. Automation is what makes that matching durable after the Award period.

Upon advancement, EcoQuest will unveil licensed production methods, implementation partners, and a confidential technical schedule under diligence. The public claim in this document is the architecture, the San Francisco use case, and the sequence by which those partners and methods enter the work.

Beginning in San Francisco is a test, not a brand claim. If the methods cannot change unit cost in one of the country's most expensive housing markets, they should not be presented as a national model. If they can, the export is the playbook, not a second claim on this Award. Boundless funds the San Francisco work.

If the layer operates as specified, the city holds more than another program. It holds a record, a pathway, a reusable method library, and, where production entities are formed, assets that can service their own capital. Talent and firms stay only if housing, food, and power are solvable. The base conditions are still too expensive. That is why the Civic rails attach to published county numbers before they attach to a workforce slogan.

• Population: California Department of Finance estimated about 842,000 residents in January 2025, down about 3,300 in one year and more than 31,000 since January 2020, about 3.6% below the start of that year. In 2024 San Francisco had the largest year-over-year drop of any California county with at least 200,000 people, while the state overall grew. (SF Chronicle, May 3, 2025.) A county estimate, not a census of who left because of rent; context, not a cause.
• Office: vacancy around 32% in Q2 2026, down from 35.4% a year earlier, after four quarters of positive absorption (2.12 million square feet in H1 2026). The top 20 tenants occupy 11.2 million square feet versus 19.5 million in 2019. Effective rents are up about 15% year over year to $66 per square foot. (Cushman & Wakefield Q2 2026 via the SF Center for Economic Development; the top-20 tenant footprint is Chronicle reporting on the same market data, April 2026.) A commercial-market signal, not a count of dark storefronts a service block can fill.
• Business base: 167,711 active registered businesses as of July 2026; 12,246 new registrations in 2025. Survival of all-time registrations: 46.2%. (SF Treasurer & Tax Collector via sfcentered.org, July 2026.) Registration is not operation, and the survival share is over all-time registrations, not a one-year failure rate.
• The constraint on the worker, not the firm: a household must earn $75.96 an hour to afford the $3,950 average asking rent (California Housing Partnership, May 2026). One-bedroom asking rent hit $3,480 in May 2026, up 17% in a year (Apartment List / Chronicle). AI tenants are expanding office footprints. The people who cook, nurse, teach, and build cannot bid against that rent.

1. 1.Their employees can live here without a $76 wage just to rent.
2. 2.Public allocations already in the city (vouchers, CalFresh, LIHEAP) could buy more stability per dollar if unit cost falls, which would cut street disorder that boards still put in the risk memo.
3. 3.San Francisco could become the reference site for a model other cities may license, which is the opposite of a city that other regions treat as a cautionary tale.

In that frame, social-service infrastructure is also an economic-development investment. A referral line alone is not industrial policy; a verified reduction in unit cost can be.

The first funded deployment would be San Francisco. Whether the approach becomes transferable depends on what is measured here. The work would build the coordination layer (matching, the ledger, expert review, and problem submission) so existing public dollars and any later production methods can be assigned at city scale. Cost reduction would still have to be demonstrated in this market. No vendor is a condition of this application.

Civil legal-aid attorneys and clinics that opt in: housing court, benefits denials, wage claims, civil-rights intake, and the humanitarian matters those offices already practice. The match is a time with an advocate, not a chatbot opinion. The present rent emergency is, among other things, a legal-aid and rental-assistance problem; an opt-in civil desk with a time on the calendar is a different object from a PDF of office hours.

Clinics and urgent slots that opt in, beside coverage navigation already authorized on the footprint. The application does not diagnose and does not hold a medical file on a public scan.

After 911 and 988 where those numbers apply: overnight capacity, advocates, and teams that are actually staffed. The rail hands a circumstance to a licensed office that opted in; it never replaces the emergency numbers.

Crankstart's published theory of change, stated in its Streets to Stability report and carried into the April 2025 roadmap, reads: make homelessness rare, brief, and one-time, through disciplined management and clear measurement. Rare means preventing the fall into homelessness. Brief means an episode that ends quickly. One-time means it does not recur. Discipline means named tools, data shared across agencies, contracts written on results, and the city using its strength in technology. The report also sets the principles for the work: focus on people who need care, lead with an equity lens, balance neighborhood needs, make the hard calls, act without waiting for the perfect answer, and pursue lasting solutions rather than short-term fixes.

This proposal is built as the software layer inside that theory. It does not replace the housing, treatment, or benefits decisions the theory requires; it is the shared record, the posted demand, and the results measurement the theory names as missing. The alignment, element by element, is stated below so a reviewer can check it rather than take it on trust.

Each alignment claim carries its own falsification test, stated later in this section: if departments will not read the ledger, the measurement claim fails; if a household still repeats intake, the coordination claim fails; if no licensed shop bids a block, the local-capacity claim fails. The alignment is offered as a mapping a reviewer can check against Crankstart's own documents, not as an endorsement from Crankstart, which this filing does not speak for.

In 2024 Crankstart's Streets to Stability report, as reported by The San Francisco Standard on October 15, 2024, mapped at least 232 providers funded by nine city agencies (the report's own text says "more than 230"; the map is not an HSH contract roster and includes groups Crankstart classified as homelessness-related, including some prevention grants; it is not a list of every place a person can get help) and described what the map showed: billions have been spent, there is no accurate account of total cost, services overlap, and a person may have to navigate fifteen to thirty providers. That navigation figure is the finding that holds. A later Crankstart roadmap, offered as suggestions rather than as city policy, stated the aim as homelessness that is rare, brief, and one-time, with disciplined management and clear measurements. The plan page names the tools it wants: streamlined intake, digital navigation, data-sharing so no one slips through, and contracting on results, and it says San Francisco should use its strength in technology. Its twelve recommendations group into three jobs: prevent, make it brief, and keep people from returning. Boundless, separately, asks for a plan that primarily benefits San Francisco and lasts. It does not prescribe homelessness software. That diagnosis is accepted here. This application is not a disagreement with it.

The limit comes next, stated plainly. The Civic Concierge does not merge those agencies or issue benefits, and it does not build homes. Crankstart's step 1 is a plan to permit 46,000 affordable homes by 2032, the affordable share of the state's 82,069-unit RHNA for 2023–2031 (press shorthand says 42,000); the City stood at about 12 percent of RHNA pace in 2025. Building those units stays with the Housing Authority, MOHCD, and builders. It is the digital experience and shared record their step 5 asks for, and the measurement layer their "clear measurements" line asks for. It sits above Coordinated Entry, 211, DAHLIA, the Housing Authority, and Human Services, and it keeps one authorized household record after the referral. Prevention rails (rental and mortgage aid, weatherization, CalFresh take-up among the 61 percent of food-insecure low-income residents not on the program) are aimed at the 9,552 first-time entries, which is the inflow the roadmap named.

Of the twelve recommendations, this application answers two directly and supports three others. It is not the housing-production plan, not a department merger, and not a new entitlement. Each feature is mapped to a numbered recommendation, not to the crisis in general.

What would falsify each claim. If departments will not read the ledger, the measurement claim fails. If a household still repeats intake, the coordination claim fails. If no licensed shop bids a block, the local-capacity claim fails. Those are the tests this filing accepts, because they are the credibility a careful funder scores.

The position, in one paragraph: this work is built for steps 2 and 5, and for the measurement and results-contracting standard in their conclusion. It is not the housing-production plan, not a department merger, and not a new entitlement. The layer is how the city uses the programs it already funds without making the household restart at each office.

The twelve are Crankstart's own recommendations: the roadmap released in April 2025 ("rare, brief, one-time," reported by the San Francisco Chronicle), not a list this application invented. They are reproduced here with what each one needs, so the mapping is checkable against that document rather than against this filing's summary of it.

Scope note: this application does not ask Boundless to fund, or to commit to, all twelve. What follows is how one standing coordination layer would facilitate each recommendation beyond what this filing is asking for: the uses that would still be open to the City, to Crankstart, or to other funders after this Award, whether or not this proposal is funded. Only the footprint, the qualifying match, the board, and the ledger are in the ask itself.

All twelve can sit on the same mechanism. The footprint shows the need. An agency signature, a contract, or a pool makes that need something a builder, a clinic, or a shop can staff. The board keeps the miss visible until someone takes it. The City still permits, licenses, and decides eligibility.

The limit on every row is the same. Without an agency or a funder behind the block, it is a map. With that backing, the twelve are one system: see the need, make it a contract, build or staff against it, and keep what worked.

That is a real path to recommendation 1, if it is stated as a viability layer and not as a permit. Crankstart's first step is a plan to permit 46,000 affordable homes by 2032, out of the state's 82,000. The bottleneck they name is zoning, a two-year permit, a cost near $1 million a unit, and no funding plan. A footprint does not rezone a parcel. It can change whether a project is financeable.

The mechanism is this. The household record shows who already qualifies, in which district, for a voucher, a rent subsidy, or a prevention payment. Agency backing turns that from a list into a commitment: a set number of households, a rent the agency will support, a term. A posted block or a pooled assistance contract turns that commitment into income a builder or operator can underwrite. A lender can size a loan against a week of signed demand instead of a hope that the lottery fills. That is how the same layer that matches a meal can sit under a bed.

The rest of the ecosystem is what turns that demand into rooms and services. Service blocks give a licensed local contractor a volume of rehab or small construction it can staff. A housing rail keeps the voucher the household already has attached to a unit that will take it, and writes back when the lease is signed, so a built unit does not sit empty. The Bounty Board is where a miss (a site with no operator, a building with vacant supportive beds) stays visible for a provider or a funder to take. Playbooks keep a method that worked so the next building does not start from a blank solicitation.

What still has to be said, or a reviewer will reject the housing claim: the City permits. The Housing Authority issues the voucher. A builder builds. The software footprints the need, and the contract or the pool is what makes that need bankable. Without the agency signature, it is a map. With the signature, it is the demand side of the 46,000.

When federal SNAP lapsed in 2025, Crankstart matched City funds with $9 million so that about 112,000 residents, in about 82,000 households, could buy food. The announcement is a public-private fund for families facing food insecurity; it is not a hot-meal program and not evidence that pooling benefits is already authorized. That partnership ran through Human Services, the San Francisco-Marin Food Bank, and GiveCard. Crankstart separately committed $10 million to the Breaking the Cycle Fund, alongside Tipping Point Community and the Charles and Helen Schwab Foundation, for interim shelter, treatment, and family prevention. It has also backed a grocery, with 18 Reasons and Bi-Rite, in the city's southeast. Those were correct emergency and neighborhood acts. They were not a standing coordination system.

Those acts treat food and crisis housing as base conditions, not as extras after a job or a degree. This proposal uses the same order: stabilize, participate, produce, own. The Civic layer is what remains after an emergency match ends: the same household connected to ongoing CalFresh, LIHEAP, vouchers, and licensed lower-cost production, rather than to a card that expires with the crisis.

In April 2025 Crankstart put $7 million into Freedom West 2.0 in the Fillmore: a mixed-income cooperative rebuild whose stated purpose is less permanent subsidy, more commercial revenue, and residents who are not displaced. Housing Security grantmaking in San Francisco and Oakland has gone to the operators who already run this city's buildings and beds (Episcopal Community Services, Bay Area Community Services, 3rd Street Youth Center and Clinic, Chinatown Community Development Center, Tenderloin Neighborhood Development Corporation, Mercy Housing, Enterprise Community Partners, the Nonprofit Housing Association of Northern California) and to policy capacity at the Terner Center. On mobility, a $14 million UCSF gift is training about 2,000 San Francisco residents for health-care jobs; Crankstart Scholarships now carry SFUSD seniors through college with advising attached; Golden State Opportunity and Code for America work the benefits-access problem this proposal generalizes.

Freedom West is the ownership end of a pathway, in a separate legal form, designed to stop needing a new public check every year. UCSF training and the scholarships are the participation rail: work and education after a household is stable enough to use them. Code for America and Golden State Opportunity are benefits access for one or two programs; the Civic footprint is that function across SNAP, vouchers, LIHEAP, and local aid. Production capital, if used, would sit in the same class as recyclable capital next to assets that are meant to amortize, never inside the application.

Read that way, this is not a request to repeat Freedom West, the SNAP match, or the UCSF training gift. Those were the right instruments for operators, emergencies, and pipelines. This is a request for the instrument those gifts could not be: a standing coordination layer, with production methods held apart, so the next emergency match is not the only way 112,000 people eat and the next capital campaign is not the only way a cooperative stays solvent.

Why those projects were fundable is not mysterious. They were specific to San Francisco. They named a number of people. They ran through an institution the city already trusted. They either prevented a fall (food, family homelessness, displacement at Freedom West) or opened a path off aid (a job, a degree, a unit that can cash-flow). They accepted measurement. They did not ask philanthropy to become a thirteenth city department. They did not promise that a new brand would retire homelessness. They promised that a particular service would work better, or that a particular asset would need less subsidy.

What is different here is the object of the work. Prior Crankstart-funded work in this city has mostly paid operators and institutions to do one job well: build or keep housing, deliver a month of food, train a cohort, get a student through college, open a benefit. This application pays for the layer those jobs sit under. A household that receives a Freedom West unit, a CalFresh allotment, an LIHEAP credit, and a UCSF training slot still has four files. The Civic Concierge is one authorized file, a qualifying match limited to this county, a sequence that does not end at the bed or the check, and a ledger that reports what the dollar bought. Production methods for farming, construction, and retrofit, held in separate entities, are how unit cost would be attacked. They are not how eligibility is decided. Upon advancement, those methods and vendors will be unveiled under diligence. That is a different object than a grant to a CDC. It is also a different risk. The honest name for that risk is early citywide software plus adoption by departments that already run Coordinated Entry, DAHLIA, vouchers, and CalFresh.

This application does not speak for Crankstart or for any organization named above. A prior grant is not a letter of intent. Those groups remain independent. This submission does not attach their letters. The point of the mapping is narrower. If the test is whether this is another platform in search of a problem, the answer is no. The problem Crankstart has already described is a fragmented provider system with weak measurement, thin prevention, and subsidy that does not retire. The Civic Concierge is the coordination layer for that problem, in this city, with production methods held in separate entities and a cost target that will be published whether or not it is hit.

Crankstart Foundation. "Breaking the cycle of homelessness in San Francisco" (Streets to Stability), 2024. streetstostability.com. Source for the principles and the rare, brief, one-time aim quoted in this section.

Lever for Change. "Crankstart and Lever for Change announce $150 million competition for the world's best ideas to improve San Francisco." August 5, 2026. https://leverforchange.org/article/press-release/announcing-boundless-san-francisco/.

Inside Philanthropy. "How to get funding from the Crankstart Foundation." Updated July 2, 2026. Housing Security program (San Francisco and Oakland); education and economic-mobility grantees.

Crankstart Foundation. Form 990 grant compilations, 2020–2024, as aggregated by philanthropy.org (San Francisco Foundation, Tipping Point Community, Tenderloin Neighborhood Development, Enterprise Community Partners, Chinatown CDC, Larkin Street, Mercy Housing, MEDA, Bay Area Community Services, and others).

Freedom West Homes Corporation. "Crankstart Awards $7 Million Grant to Freedom West Community Development Corporation." PR Newswire, April 16, 2025.

Angst, Maggie. "S.F. Mayor Lurie unveils first big haul of private funds to address homelessness." San Francisco Chronicle, May 8, 2025. Breaking the Cycle Fund: Tipping Point $11M, Schwab $10M, Crankstart $10M.

San Francisco Human Services Agency. "Mayor Lurie and Crankstart Partner to Provide Funding to Support San Francisco's Families Facing Food Insecurity." October 29, 2025. ~$18 million public-private fund for families facing food insecurity (not a hot-meal program); Crankstart $9 million; ~112,000 residents / ~82,000 households.

Sjostedt, David. "See the vast web of homelessness nonprofits in one baffling infographic." The San Francisco Standard, October 15, 2024. Crankstart map of at least 232 providers and nine agencies; 15-to-30 provider navigation line; 12-step framing including prevention; the plan figure is 46,000 affordable homes permitted by 2032, the affordable share of the state's 82,069-unit RHNA ("42,000" is press shorthand).

Graff, Amy / San Francisco Chronicle. "Lurie ally releases roadmap to address S.F.'s homelessness crisis." April 1, 2025. Crankstart roadmap: rare, brief, one-time; disciplined management and clear measurements.

Mission Local. "Michael Moritz's foundation gives $3.4M to S.F. immigrant defense." June 23, 2025. Context on Crankstart gifts to City departments since 2021.

Talerico, Kate. "Barbara Lee lands $9.3 million philanthropic boost for Oakland anti-dumping efforts." San Francisco Chronicle, June 11, 2026. Also reporting Crankstart's $9 million SF food assistance and Embarcadero Plaza commitment.

UCSF announcement of a $14 million Crankstart gift to train approximately 2,000 San Francisco residents for health-care careers over five years. Cited via Inside Philanthropy, July 2026.

Advancement triggers a defined sequence. The items below are commitments of that sequence, not a list of deficiencies in this filing.

Partners and methods. Licensed production methods, implementation partners, and the confidential technical schedule will be unveiled under diligence. Neighborhood organizations already serving the pilot district will be named as navigator subgrantees. City-system interfaces will be opened with HSH, SFHA, SFHSA, MOHCD, and 211 as those departments agree.

Pilot. One San Francisco district. Twelve months. Baselines taken from published HSH and SFHSA reports: first-time entries, exits to housing, CalFresh reach, voucher lease-up time. Match-quality gates control any expansion beyond the district.

Operating team. A San Francisco managing director who already works these systems. An office in the pilot district. An independent evaluator retained in the city. Financial control and charitable-use documentation required of a for-profit Lead, prepared for expenditure responsibility.

Measurement. A 30 percent reduction below market cost remains a planning target drawn from comparable licensed production elsewhere, not a result recorded in this city. The delivered ratio will be published for housing, food, and energy even when it is 1.00. No portal field will treat the target as a completed result.

Scope. Primary benefit remains the City and County of San Francisco. National or later-standard language describes where the architecture can travel after this city has a measured ledger. It does not relocate the work.

Production. First housing, food, and energy sites, if formed, will sit in separate San Francisco entities. Title will not sit in the application. Public benefits will continue to be disbursed by the agencies that own them.

1. 1.People unsheltered or in shelter (7,973 on PIT night; 23,561 across a year).
2. 2.Families: the cohort the 2024 count showed rising fastest: a 94 percent increase, with 67 percent of those families sheltered and 90 percent of unsheltered families in vehicles (City release on the 2024 PIT, May 2024).
3. 3.Veterans: 327 remaining; finish the job.
4. 4.Elders, disabled residents, children on the same footprints.
5. 5.CalFresh and voucher households not yet in crisis, to cut the 9,552 first-time entries.

Point-in-Time and system volume

A note on the two volume measures in this table. The night-count rows and the system rows measure different things. First-time entries, exits, and the three-in-for-one-out ratio come from HSH system reports (ONE System and the Needs Assessment), not from the Point-in-Time count. A night count can fall while first-time entries stay high, so the two are not presented here as confirming each other.

A note on assessments. A Coordinated Entry assessment is a scored questionnaire, 0 to 160, used to prioritize a scarce queue. A high score is still not a unit; the Controller's December 2024 shelter assessment found most people assessed through Coordinated Entry do not qualify for permanent supportive housing; and a household that does not score high enough cannot request its own review; a case manager has to. The gap the documents support is assessments without placements, not a missing assessment form. The layer does not replace that score or move anyone up the queue; it can show whether a household has already been assessed, whether it holds a referral status, and whether a lease was signed.

A note on provider counts. Directories of community programs can list thousands of entries for San Francisco ZIPs, because they count programs and locations, not unique city contractors. This monograph uses Crankstart's 2024 map of at least 232 homelessness-related providers funded by nine agencies (funded providers, not directory listings). The report's own text says "more than 230." The set is not an HSH contract roster; it includes groups Crankstart classified as homelessness-related, including some prevention grants. It is not a list of every place a person can get help. The finding that holds for a household is the navigation burden: fifteen to thirty providers.

What one department spends

HSH's budget is a department budget, not total city cost. Crankstart's own point is that hospital, court, and jail costs sit outside this number, so do not read it as "what San Francisco spends on homelessness."

Cost of a night vs a home, for families (city and advocate figures)

HSH notes the comparison is imperfect: people stay long-term in housing and (if the system works) shorter-term in shelter. The Civic layer's job is to make the cheaper durable option the one that actually happens.

UCSF Benioff Homelessness and Housing Initiative, California Statewide Study of People Experiencing Homelessness (CASPEH), interviews 2021–22, published 2023, the largest representative U.S. homelessness study since the mid-1990s (~3,200 questionnaires, 365 in-depth interviews):

• Median monthly household income in the six months before homelessness: $960.
• 90% last lost housing in California; 75% in the same county where they were counted. This is not a migration story.
• Median length of homelessness: 22 months.
• 47% of adults were 50 or older. Black and Native people are sharply overrepresented.
• 19% entered from an institution; 49% from a non-leaseholder situation (doubled up); 32% from a lease.
• Leaseholders: median income $1,400, median housing cost $700; 66% of those who paid rent were rent-burdened. Most common leaseholder reason for leaving: loss of income (21%).
• Non-leaseholders had a median of one day of notice. Leaseholders had a median of ten.
• Most participants said a rental subsidy or one-time financial help would have prevented homelessness.

Source: Kushel et al., Toward a New Understanding, UCSF BHHI 2023; UCSF news June 19, 2023; later BHHI health follow-on November 18, 2025.

That finding is why the Civic housing rail starts before the street (first-time buyer, down payment, rental aid, mortgage aid) and why the footprint includes households still housed on CalFresh and vouchers. Prevention is the 9,552 first-time entries.

This proposal does not argue that San Francisco should spend less. It argues that $832.7 million at HSH, $441.6 million in voucher authority, and approximately $225 million in CalFresh benefits should purchase exits and nutrition rather than another year at the same unit cost.

Named software already handles search, referral, and the homelessness record. None of it shows unmet weekly need or organizes the work that would meet it.

No product on this list lets a household opt a benefit into a posted block and keep the benefit in their name, lets a licensed shop bid that week, and writes the finished work to a City view by district. That is the part that is not already software.

A harder look turns up more overlap than 211 alone, and it sharpens what this filing must not do. Lookup and referral are covered: 211 does live information and referral, including shelter; findhelp is under city contract through 2029; the SF-Marin Food Bank already has a pantry locator; Human Services lists free-meal sites; DAHLIA lists affordable housing and runs the lottery. None of these show unnamed neighborhood demand or staff a week of work. Housing intake is covered, and it is a scored queue: Coordinated Entry, in the ONE System, is the intake point: it assesses, scores, and refers. A second assessment would collide with that system. The housing rail is only new if it shows units that will take a voucher the household already has, and whether a lease was signed; another listing of affordable buildings would only repeat DAHLIA. Hot meals for some households already exist: the CalFresh Restaurant Meals Program has been statewide since 2021, and in San Francisco an elder, a person with a disability, their spouse, or someone experiencing homelessness can already buy a prepared meal with an EBT card at participating restaurants. A general "we invented hot meals" claim would be wrong. What is not in that program is a posted service block, a route, or a prepared-meal option for a household that does not meet those rules; those still need local funds or a later rule.

Closed-loop referral is a known product, and it often fails in practice. CalAIM requires Medi-Cal plans to close the loop on community referrals. North Carolina's NCCARE360 showed the limit: when organizations were paid to resolve food referrals, 88 percent were confirmed; after the pay stopped, confirmation fell to 30 percent. The software was the same. Payment was the difference. A ledger that only asks a nonprofit to log a referral is not new, and it will not hold without a reason to close the ticket.

Still not found:

• No City system publishes this week's unmet need by type and neighborhood with no name and no face.
• No system groups those tickets into one route a kitchen can run.
• No system lets a household opt a benefit into a posted block, keep the benefit in their name, and have a licensed shop bid the week.
• No public screen counts finished neighborhood meals, rides, and leases across agencies as completions rather than enrollments.
• No board keeps the unfilled ticket open for a solver, a nonprofit, or a funder after the referral closes.

The sentence that survives this look: do not build another directory, another assessment, or another hot-meal right for people who already have one. Build the posted week of work, the opt-in tally, and the public record of what was finished.

Existing instruments function as an intake point, a directory, a lottery, or a funding stream. None currently maintains an authorized, cross-program picture of one household's public lines, matches that household only to programs it qualifies for in San Francisco County, sequences those matches as stabilize → participate → produce → own, or reports delivered value and exits. Prior philanthropy in this city has mostly paid one operator to do one job well. This work pays for the layer those jobs sit under. A household that receives a voucher, a CalFresh allotment, an energy credit, and a training slot still has four files. Civic Concierge is specified as one file.

Benefit to the city, stated as implications. First, public dollars already appropriated are used by the household that can use them. Second, first-time entries meet prevention and housing rails before a new shelter bed is the only remaining option. Third, when a published target is missed, the Bounty Board and Brain Trusts produce a method the city can reuse and license, so giving funds a solved procedure rather than the same miss twice. Fourth, small allotments may be pooled, with consent, in separate entities that can buy or produce at a scale one stipend cannot. Fifth, if unit cost later falls, the same public dollar models as more goods, a planning target, published even when the ratio is 1.00. Sixth, where surplus appears, a defined share returns to the giving mechanism. That is how city giving is specified to yield a dividend without the application owning the building or claiming the benefit.

Nothing in current San Francisco practice does all of the following at once: (1) footprints an individual across housing, nutrition, energy, and health allocations; (2) filters the city's real inventory by qualification and jurisdiction; (3) writes a pathway with an ending; (4) counts delivered value and exits by community category; (5) lets small public stipends opt into wholesale production; (6) aims licensed methods at unit cost as a published target; (7) runs the same nine pillars as expert Civic Brain Trusts feeding a Bounty Board. That combination is the product.

Coordinated entry and HMIS, nationally, reduce some repeat-crisis use when the match is real. HUD requires Continuums of Care to operate coordinated entry; San Francisco already does. We are not inventing a new intake point. We are putting the rest of the household's public dollars on the same page as the intake point the city already runs.

Housing First retention literature shows people stay housed when the unit is real and services follow. CASPEH found most people named money and rent, not a desire to remain unhoused. Our housing rail takes that finding literally.

SNAP access studies (USDA, GAO, CBPP) show eligible households leave benefits unused when the process is fragmented. SF's own Food Security Task Force number (only 39% of food-insecure residents under 200% FPL on CalFresh) is the local proof.

Weatherization and community-supported agriculture show public or pooled dollars can buy more physical goods when unit cost falls. We will not import those percentages into San Francisco as facts. We will measure.

Voucher lease-up research (Furman / Strochak) shows time and landlord participation, not authorized dollars, are the constraint in tight markets. SFHA's 99.8% budget utilization (HUD dashboard via the Budget and Legislative Analyst, June 2026; includes project-based vouchers) next to 20,341 households without an affordable home is the local proof.

City and County of San Francisco, 2026 Point-in-Time Count preliminary results and Mayor's Office release, May 12, 2026.

Department of Homelessness and Supportive Housing, Data Report slides and ONE System figures, May 2026; 2026 HIC and PIT results deck.

Our City, Our Home Oversight Committee, 2025 Homelessness Needs Assessment, adopted January 22, 2026.

HSH Budget Fiscal Year 2026 to 2028, adopted July 29, 2026, sf.gov.

HSH Budget Overview presentation, FY2025–26 service-area dollars, assessments, encounters.

San Francisco Civil Grand Jury, At Scale, At Risk: Upgrading Data and Oversight to Improve Homelessness Response, 2026.

Budget and Legislative Analyst, Financing Affordable Housing, report to Supervisor Chen, June 5, 2026.

California Housing Partnership, San Francisco County Affordable Housing Needs Report, May 2026.

SF Chronicle / Apartment List rent series, June 25, 2026.

HUD Housing Choice Voucher dashboard figures as cited by the BLA, year-end 2025.

San Francisco Housing Authority, Housing Choice Voucher participant guidance and EHV wind-down, sfha.org.

San Francisco Food Security Task Force, 2026 Recommendations (CHIS 2024 food insecurity; SFHSA CalFresh December 2025).

SF Human Services Agency, CalFresh eligibility and FY2026 allotment tables, sfhsa.org.

SFDPH–HSA Budget Impact Memo on federal cuts, November 2025.

ABC7 San Francisco, 17,000 residents facing SNAP cuts, September 2, 2026.

Center on Budget and Policy Priorities, SNAP Tracker, August 26, 2026.

Legislative Analyst's Office, The 2026–27 Budget: Food Assistance Programs, February 17, 2026.

USDA SNAP maximum allotments, FY2026 COLA effective October 1, 2025.

UCSF Benioff Homelessness and Housing Initiative, Toward a New Understanding (CASPEH), 2023; UCSF news June 19, 2023; Toward Thriving health report November 18, 2025.

United Way Bay Area / Keisha Browder, "The Bay Area's 211 helpline is in crisis," SF Chronicle Open Forum, December 28, 2025.

KQED, Civil Grand Jury coverage of exits −14.3%, June 23, 2026.

The Frisc, family shelter vs housing costs, July 1, 2025.

USC Homelessness Policy Research Institute, California budget brief, June 7, 2026.

Public Policy Institute of California, Homelessness and Drug Use in California, August 26, 2026.

Hoover Institution, Ohanian on statewide spend vs counts, February 2026.

Boundless San Francisco Rules, Evaluation Rubric, Timeline, and Organization Readiness Tool, boundless-sf.leverforchange.org, 2026.

CalAIM community supports network referral requirements, California Department of Health Care Services; NCCARE360 evaluation finding: 88 percent of food referrals confirmed while organizations were paid to resolve them, 30 percent after payment stopped.

Counsel and reviewers will ask four questions. They are fair.

First, is this findhelp with a new name? No. A directory answers "what programs exist." A footprint answers "what this household already holds and may use." A block answers "how much licensed work this district will pay for this month." The first is a list. The second and third are a commercial and civic instrument. If implementation delivered only the list, the project would have failed its own test. A qualifying-match screen that only lists programs is redundant with 211 and with findhelp. The housing rail is only new if it shows units that will take a voucher the household already has and whether a lease was signed; if it is another listing of affordable buildings, DAHLIA already does that. The line that holds up if someone already knows findhelp: we are not another list of programs; the City is already buying that. The sentence that is not redundant: we show this week's need, group it into work a licensed local provider can staff, and record what was finished.

Second, can allotments be pooled or stretched without breaking program rules? Not by assertion. Nutrition, housing, energy, and care each have their own statute. This filing treats the pathway as the first product: opt-in, match, vendor standard, ledger. A plate, a visit, or a seat that uses an existing allotment runs only where present law already allows it, or where a later waiver is granted and attached to a pathway that already works. Local match and spare capacity in a kitchen that already staffs a line are the first stretch tools. No change to federal nutrition law is claimed as already won.

Third, is a household footprint surveillance? The public view is blocks, not people. Consent is specific. The file is not sold. Category tags used for reporting are the tags departments already publish. If a household will not authorize a richer file, they still receive a public match at the depth they choose. That is a product rule, not a slogan.

Fourth, will small licensed operators actually be able to bid? Only if the City's procurement for a block is written to the size of a neighborhood shop, not only to the size of a prime contractor. The first blocks named under diligence would be sized for that test: one nutrition block and one care block, each with a published standard, a bidding rule, and a public metric. If only incumbents can clear the paperwork, the board has reproduced the maze. That failure would be visible on the ledger.

Boundless asks whether an approach will shift conditions for people who live and work in San Francisco, and whether the shift can last. The condition this filing takes as given is the fragmented provider system already documented by the sponsor: many paid providers, nine agencies, almost no unused voucher authority, benefits that still miss people who qualify, and shops that cannot see demand until it is too late to staff it.

The shift offered is not a new department and not a replacement for CalFresh, Coordinated Entry, DAHLIA, or Public Health. It is a coordination layer that sits beside those systems: one authorized record where the household permits it, qualifying matches instead of a new intake, posted blocks instead of guessed walk-up, ordinary license and bond instead of a parallel code, and a ledger that shows units delivered and whether any extra stayed here.

That is all this public filing claims. Partners, operators, and the first measured blocks would be unveiled under diligence if this entry advances. Until then, the honest position is the one above: the doors exist; the board does not; usefulness lives or dies on whether the board can be made lawful and small enough for a neighborhood to use.

Condition taken as given. San Francisco already pays for the programs. Crankstart's 2024 map counted at least 232 homelessness-related providers funded by nine agencies, with overlap and fragmentation that may send a household through fifteen to thirty providers. A listing and a guide help a person reach one service. They do not change the structure that leaves a licensed kitchen guessing at volume.

What is supposed to shift is the opportunity structure around existing statutes: one authorized footprint instead of a repeated intake; a posted block of qualifying work instead of walk-up demand; a ledger of units delivered instead of another annual appropriation whose result cannot be read across agencies. Direct service exists so that structure can be seen and tested. It is not the end of the theory.

Theory of change, in one chain: if a household may authorize a picture of what they already receive and may qualify for in this county, and if need that can be aggregated without naming a person is posted as a block, and if a provider that already holds the ordinary license, bond, and insurance may bid, and if delivery is written to a public ledger, then two conditions change. The household stops starting over. A neighborhood operator can staff a line against predicted work. The City can see whether a dollar bought a unit, a stretch, or only another referral.

Proportion. This filing specifies a coordination layer and the first two blocks against which that layer would be tested. Ambition that outruns that instrument would not be proportionate. Ambition that stops at a better directory would not be transformative. The standard this filing sets for itself is a working service board, not a better search page.

Context already used: Coordinated Entry, DAHLIA, voucher utilization, CalFresh and HSA, Home by the Bay, DPH contracted inventory, 211 and the City's contracted directory, the 232-provider map. Those are the services the Civic layer sits beside.

What is not claimed as already in hand: no data-use agreement, no named kitchen or clinic, no exclusive technology partner. That is the viability gap. The path below closes it without waiting on an award for work that does not require one.

1. 1.Months 0–3, with or without an award. Stand up the lead charitable affiliate that can receive restricted funds and contract with the City. Publish the specification against public inventory only (DAHLIA, published program rules, directory listings as a catalog). Accept household-volunteered proofs. No agency feed required for a shallow match.
2. 2.Months 3–6. Design two block templates with program counsel (one nutrition, one care), including the license each bidder must already hold. Identify the class of operator (permitted kitchen, contracted clinic, licensed child-care site) without naming a vendor. Open conversations with HSA, HSH, DPH, and the City Attorney on procurement sized to a neighborhood shop.
3. 3.Months 6–12, if advanced. Name the first two measured blocks under diligence. Execute a limited data-use agreement. Unveil operators then. If only incumbents can clear the paperwork, the board has reproduced the present fragmentation; that failure would be visible on the ledger.

This filing can show that the problem is documented here and that adjacent tools exist and where they stop. It cannot show EcoQuest Civic meals, appointments, or prevention visits completed in San Francisco. Those units do not exist yet. Treating modeled stretch as a local result would be a misrepresentation.

Problem evidence: At least 232 providers mapped by Crankstart in 2024, not a contract roster (SF Standard, October 15, 2024; the 15-to-30 navigation burden is the finding that holds); Home by the Bay gap; voucher budget utilization near the ceiling, not lease-up; CalFresh still below the eligible population; a City-contracted directory already in place for older adults and people with disabilities.

Adjacent tools, and the line they do not cross: 211 and the directory answer what programs exist; Coordinated Entry ranks a subset of homeless households; benefits-access groups file for one or two programs; closed-loop referral can raise completed handoffs and has not, on its own, been shown to change housing exits or food security at city scale. None of those tools post a block a licensed neighborhood operator can bid.

Analogs this theory is allowed to use: known volume lowers meal unit cost; spare capacity in an already-staffed line absorbs off-peak work; California's Restaurant Meals Program already lets a defined SNAP population buy prepared meals; an intake point without a ledger of completed steps reproduces the same fragmentation in software.

Year-one results, if advanced: blocks posted; eligible bids from operators that are not the usual prime; units delivered; share who completed the next matched step; remainder disposition; time from footprint to first qualifying match. An independent evaluator would see the same ledger. Those numbers do not exist today.

Institutional adoption: eligibility stays with the agencies. Policy: a procurement template sized to a neighborhood shop is what lets the board outlive a grant. Community ownership: remainder follows a City rule; the playbook stays with the jurisdiction. Financial: after a grant the layer is meant to be a city-procured utility or a nonprofit contract, paid as infrastructure. Production assets, if any, sit elsewhere and are not the durability plan for the software.

Five-year sketch. Year 1: two measured blocks, public metrics, charitable affiliate contracting, evaluator on the ledger. Years 2–3: reuse a template only if year-one units cleared the standard; remainder rule as administrative practice. Years 4–5: the board is written into a city contract as a replaceable utility, or it sunsets. Playbooks public. EcoQuest, if it continues, is an operator under contract. Lasting benefit is the template and the ledger, not a vendor lock. If year five arrives and the only durable object is a branded app, the project will have failed durability by its own standard.

• Ambition: opportunity structure (footprint, block, ledger), not a new department.
• Viability: specific city context; small first blocks; presence can start on public inventory; award vehicle is an eligible charity, not the LLC.
• Evidence: early-stage. Problem documented, analogs bounded, no local Civic units claimed.
• Durability: the City keeps the rule and the playbook; year five is a utility or a sunset.

Anything beyond these statements is not claimed in this filing.

A reviewer is asked to accept only the following: existing services remain; a directory is not the missing instrument; two lawful blocks are a proportionate first examination of a coordination layer; SNAP allotments are not taken from the household; what is published is the block rather than the person; the award is held by an eligible charity; and the fifth year can belong to the City.

A page that requires any further belief should be removed from the packet.

1. 1.Lever for Change / Crankstart. Boundless San Francisco. Competition website, rules, evaluation process, and four-criterion rubric (Transformative Ambition, Viability, Evidence of Results, Durability). https://boundless-sf.leverforchange.org.
2. 2.Lever for Change. Boundless San Francisco Application Toolkit and Organization Readiness Tool. 2026.
3. 3.Lever for Change. Boundless San Francisco Competition Rules, including 75 percent City and County of San Francisco primary-benefit requirement, 25 percent broader Bay Area cap, Award range as published by the Competition, below-market capital option, 29 percent indirect-cost cap, expenditure responsibility for non-501(c)(3) Leads, and prohibition on Award funds purchasing crypto-tokens. 2026.

1. 4.City and County of San Francisco, Department of Homelessness and Supportive Housing (HSH) and Mayor's Office. 2026 Point-in-Time Count preliminary results (count night scheduled January 29, 2026). May 12, 2026. Figures used: 7,973 people counted (−4.2% from 8,323 in 2024); 3,400 unsheltered (−22% from 4,354); sheltered share 57% (from 47.7%); families 465 (2024: 405); youth under 600 (−54% vs. 2024); veterans 327 (−44% from 585).
2. 5.City and County of San Francisco. 2024 Point-in-Time Count, night of January 30, 2024, and Mayor's release, May 2024. Figures used: 8,323 total (+7% from 7,754 in 2022); unsheltered 4,354 (−1%); sheltered 3,969; tents/sidewalks 2,912 (−13%); families 405 (+94%), 67% of them sheltered and 90% of unsheltered families in vehicles; sheltered share 48% (36% in 2019). Also: more than 20,000 people seek homeless services over a full year (SF.gov); a night count is not annual demand; first-time entries, exits, and the inflow-vs-exits ratio come from HSH system reports, not from the night count.
3. 6.San Francisco Department of Homelessness and Supportive Housing. Data Report / ONE System figures and 2026 Housing Inventory Count and Point-in-Time results deck. May 2026. Figures used: 23,561 people served in the Homelessness Response System in FY2024–25; 9,552 first-time entries; 15,768 people in permanent supportive or other permanent housing on PIT night.
4. 7.San Francisco Department of Homelessness and Supportive Housing. Home by the Bay: An Equity-Driven Plan to Prevent and End Homelessness, 2023–2028. Cited as context the Civic layer would sit beside, not as evaluation of it.
5. 8.Tipping Point Community, Charles and Helen Schwab Foundation, and Crankstart. Breaking the Cycle Fund. Announced May 2025, about $37.5 million at launch, for interim shelter, treatment, and family prevention. Cited as context the Civic layer would sit beside, not as evaluation of it.
6. 9.Our City, Our Home Oversight Committee. 2025 Homelessness Needs Assessment. Adopted January 22, 2026. Figure used: approximately three people enter homelessness for every one who exits.
7. 10.San Francisco Department of Homelessness and Supportive Housing. Budget, Fiscal Year 2026 to 2028. Adopted July 29, 2026. https://www.sf.gov. Figure used: $832.7 million HSH budget (FY2026–27), a department budget, not total city cost; hospital, court, and jail costs sit outside it.
8. 11.San Francisco Department of Homelessness and Supportive Housing. Budget Overview presentation, FY2025–26. Figure used: Coordinated Entry funding (approximately $5.9 million). Assessment volume: the Budget Overview's 12,000+ figure is an assessment count tied to that presentation, not a housing count; the last on-record official figure is 10,314 Access Points assessments in the year ending March 2023, and a January 2026 account citing HSH reported more than 19,000 people active in Coordinated Entry in the prior year.
9. 12.City and County of San Francisco, Office of the Controller. Shelter assessment. December 2024. Finding used: most people assessed through Coordinated Entry do not qualify for permanent supportive housing.
10. 13.El Tecolote. Family rental-support assessment figures, July 2024 to May 2025: 1,826 assessed; 30 placed.
11. 14.San Francisco Civil Grand Jury. At Scale, At Risk: Upgrading Data and Oversight to Improve Homelessness Response. 2026. Figures used: exits to stable housing down 14.3% year over year; oversight context beside approximately $487 million in nonprofit contracts.
12. 15.City and County of San Francisco. Adult shelter waitlist snapshot. February 2026. Figure used: 424 people.

1. 16.San Francisco Housing Authority. Housing Choice Voucher participant guidance and Emergency Housing Voucher wind-down. https://www.sfha.org. Figures used: 15,708 leased vouchers; $441.6 million in budget authority; 99.8% authority utilization (budget utilization, including project-based vouchers; monthly SFHA reports show unit-lease rates in the high 90s on a smaller allocated-voucher base).
2. 17.U.S. Department of Housing and Urban Development. Housing Choice Voucher dashboard figures as cited by the San Francisco Budget and Legislative Analyst, year-end 2025.
3. 18.Budget and Legislative Analyst, City and County of San Francisco. Financing Affordable Housing. Report to Supervisor Chen. June 5, 2026.
4. 19.California Housing Partnership. San Francisco County Affordable Housing Needs Report. May 2026. Figures used: 20,341 low-income renter households without an affordable home; 65% of extremely low-income households paying more than half of income on housing; $75.96 hourly wage required to afford $3,950 average asking rent (market-asking figures for one month, a price signal, not a count of who is cost-burdened).
5. 20.San Francisco Chronicle, citing Apartment List. One-bedroom asking-rent series. June 25, 2026. Figure used: $3,480 one-bedroom asking rent in May 2026 (+17% year over year).
6. 21.Mayor's Office of Housing and Community Development (MOHCD). DAHLIA San Francisco Housing Portal. Lottery and below-market listing system referenced as existing inventory, not replaced by this proposal. https://housing.sfgov.org.

1. 22.U.S. Department of Agriculture, Food and Nutrition Service. SNAP maximum allotments, FY2026 cost-of-living adjustment effective October 1, 2025. Figures used: 48 states and D.C. maximums: 1 person $298; 2 people $546; 3 people $785; 4 people $994 per month (issuance figures, not meals served).
2. 23.San Francisco Human Services Agency. CalFresh eligibility and FY2026 allotment tables. https://www.sfhsa.org. Figure used: approximately 111,700 San Franciscans reached by CalFresh (also styled ~110,000 / 111,700 in city materials).
3. 24.San Francisco Food Security Task Force. 2026 Recommendations. Citing California Health Interview Survey (CHIS) 2024. Figures used: food insecurity among residents under 200% of federal poverty level; 39% of those residents receiving CalFresh.
4. 25.San Francisco Department of Public Health and Human Services Agency. Budget Impact Memo on federal SNAP cuts. November 2025. Figures used: 21,080 ABAWDs at risk if they do not meet work rules; almost 20,000 people projected to lose CalFresh by the end of 2027; approximately $6 million per month local grocery-spend drop; 300 grocery jobs plus ~600 related jobs supported by CalFresh spend (citing National Grocers Association).
5. 26.ABC7 San Francisco. Reporting on residents facing SNAP / CalFresh work-requirement cuts. September 2, 2026. Figure used: approximately 17,000 San Francisco residents at risk.
6. 27.Center on Budget and Policy Priorities. SNAP Tracker. August 26, 2026. Figures used: California CalFresh participation to May 2026, −333,062 people (−6%), children −156,440 (−9%); national SNAP participation drop July 2025–May 2026, approximately 5 million people (−12%); 2025 reconciliation-law SNAP cut of approximately $187 billion through 2034 (Congressional Budget Office, as cited by CBPP and the SF Food Security Task Force).
7. 28.Legislative Analyst's Office, State of California. The 2026–27 Budget: Food Assistance Programs. February 17, 2026. Figures used: proposed CalFresh + CFAP $15.5 billion; caseload 3.22 million households (−2%).
8. 29.Congressional Budget Office. Estimates of federal SNAP reductions in the 2025 reconciliation law, through 2034. As cited by CBPP and the San Francisco Food Security Task Force.

1. 30.Kushel, Margot, et al. Toward a New Understanding: The California Statewide Study of People Experiencing Homelessness (CASPEH). UCSF Benioff Homelessness and Housing Initiative. Interviews 2021–22; published 2023. UCSF news June 19, 2023; later BHHI health follow-on November 18, 2025. Figures used: ~3,200 questionnaires and 365 in-depth interviews; median monthly household income $960 in the six months before homelessness; 90% last lost housing in California and 75% in the same county; median length of homelessness 22 months; 47% of adults 50 or older; 19% entered from an institution, 49% from a non-leaseholder situation, 32% from a lease; leaseholder median income $1,400 and median housing cost $700; 66% of those who paid rent were rent-burdened; most common leaseholder reason for leaving was loss of income (21%); non-leaseholders median one day of notice, leaseholders median ten; most participants said a rental subsidy or one-time financial help would have prevented homelessness.
2. 31.Strochak, Sarah, et al. (Furman Center / related voucher lease-up research). Housing Choice Voucher lease-up timing in tight markets. Figures used in text: national lease-up of 31% at 60 days and 61% at 180 days, cited to show that time and unit availability, not authorized dollars, bind tight markets.

1. 32.Ohanian, Lee. Analysis of California homelessness spending 2019–early 2024 versus official count change. Hoover Institution. February 2026, citing state figures. Figures used: approximately $24 billion spent while the official count rose 23%, from about 151,000 to 186,000; implied state spend per additional unhoused person over five years on the order of $160,000 (ratio as presented by the author).
2. 33.Henwood, Benjamin. Policy brief on California General Fund homelessness spend. USC Homelessness Policy Research Institute. June 7, 2026. Figure used: approximately $1.5 billion of a $321 billion General Fund (0.47%) in FY2025–26. The share is of the state General Fund for that fiscal year; a department budget and a General Fund line are not the same base.

1. 34.California Department of Finance. Population estimates. Reported by Christian Leonard, "S.F.'s population drops again as city struggles to retain residents," San Francisco Chronicle, May 3, 2025. Figures used: about 842,000 residents as of January 2025; −3,300 in one year; more than 31,000 below January 2020 (−3.6%); largest year-over-year drop among California counties with at least 200,000 people in 2024. A county estimate, not a census of who left because of rent; context, not a cause.
2. 35.San Francisco Center for Economic Development / Cushman & Wakefield. San Francisco office market, Q2 2026. https://www.sfcentered.org/data. Figures used: vacancy 32.0% (from 35.4% a year earlier); +2.12 million square feet net absorption in H1 2026; four straight quarters of positive absorption; effective rents up about 15% year over year to $66 per square foot.
3. 36.Devulapalli, Harsha / San Francisco Chronicle. "These companies have the biggest office footprints in S.F." April 27, 2026. Source note: Cushman & Wakefield and Chronicle reporting. Figures used: top 20 tenants 11.2 million square feet in 2026 versus 19.5 million in 2019 (Chronicle reporting on the same market data). A commercial-market signal, not a count of dark storefronts a service block can fill.
4. 37.Office of the Treasurer & Tax Collector, City and County of San Francisco. Registered-business counts as presented by the San Francisco Center for Economic Development, July 2026. Figures used: 167,711 active registered businesses; 12,246 new registrations in 2025; 46.2% survival share of all-time registrations. Registration is not operation, and the survival share is not a one-year failure rate.
5. 38.City and County of San Francisco. Five-Year Financial Plan, FY2025–26 through FY2029–30. December 2024. Context for post-pandemic office attendance and population loss.
6. 39.U.S. Bureau of Labor Statistics. San Francisco–Oakland–Fremont, CA Economy at a Glance. Data extracted September 9, 2026. https://www.bls.gov/eag/eag.ca_sanfrancisco_msa.htm. Labor-force context only; not used as a homelessness count.

1. 40.United Way Bay Area / 211 Bay Area (or successor information-and-referral operator). 211 referral service for the San Francisco Bay Area.
2. 41.San Francisco Department of Homelessness and Supportive Housing. Coordinated Entry. Assessment and match to homeless-housing opportunities.
3. 42.Mayor's Office of Housing and Community Development. DAHLIA San Francisco Housing Portal.
4. 43.San Francisco Housing Authority. Housing Choice Voucher program administration.
5. 44.San Francisco Human Services Agency. CalFresh, Medi-Cal eligibility administration.
6. 45.U.S. Department of Health and Human Services, Low Income Home Energy Assistance Program (LIHEAP), as administered locally.
7. 46.U.S. Department of Agriculture. Special Supplemental Nutrition Program for Women, Infants, and Children (WIC).

1. 47.EcoQuest. Civic Concierge planning model. Ten-year public-value examples, if published later, will be models only: matching and utilization; production in San Francisco; stretch only if a verified production-cost reduction and City adoption of the ledger. No dollar target is stated as a request in this draft.
2. 48.Comparable U.S. production projects (applicant-reported, unnamed in this public filing). Some operating projects have reported savings of up to 30 percent versus conventional cost. Cited only as analogous evidence for the 30 percent planning target, not as San Francisco results. Documentation would be provided in confidence if the Competition requests it after an Award decision.
3. 49.EcoQuest. SNAP value-stretch illustration. If a 30 percent under-cost production target were later verified in San Francisco, a $298 one-person maximum SNAP allotment would model as $426 of food ($298 ÷ 0.70 ≈ $426). Stated in the text as a planning target only.
4. 50.EcoQuest. Platform architecture for the dual-layer design, nine domains, Brain Trusts, Bounty Board, and legal separations. Cited only for system design. Not a source for San Francisco Point-in-Time, budget, or caseload figures.
5. 51.Access note. City and state documents listed above were the public versions available at the time of drafting (May–September 2026). If a department later revises a preliminary 2026 Point-in-Time table, the proposal will follow the official revision. URLs that sit behind a department CMS may change; titles and dates are the durable citation.