GiveBetter x SF

Code Tenderloin

Paid peer outreach, employment support and timely overdose response

Research time: ~17 min on GPT-5.6 Sol Medium + ~15 min on GPT-6.1 Sol
  • v1: ~17 min on GPT-5.6 Sol Medium; estimated before tracking.
  • Source audit — checked claims, assumptions and calculations.
  • Research — reviewed programs, finances and impact evidence.
  • Modeling — estimated costs, QALYs and uncertainty.

Updated: 2026-10-01.

Donate

Summary

What do they do? Code Tenderloin provides community-based support in San Francisco, including employment, youth and reentry services. Its ambassador work includes peer contact and help during emergencies. Our estimate combines additional timely overdose response with net resources from paid peer work, retaining the cost of other activities. More

We include Code Tenderloin among our top-ten research candidates because of its:

  • Local peer relationships that may reach people conventional services miss.
  • Paid on-the-job training that may improve workers’ resources relative to their alternatives.
  • A potentially important role in timely overdose response, alongside broader community support.

Our main reservations about Code Tenderloin are:

  • The complete worker-hour cost, paid tenure, alternative earnings and allocation of new funding are assumptions.
  • Additional high-risk reach and the change in emergency outcomes have not been measured.
  • Current accounts, public funding and historical sponsor arrangements need reconciliation before estimating extra capacity.

What do you get for your dollar?

GiveBetter sees a potentially valuable mechanism in having a trusted person nearby who can recognize and respond to an overdose. An advertised ambassador wage of about $27 an hour anchors our assumed $60 complete worker-hour cost. Six worker-hours per offered person-year give an assumed $360 delivery package; after 50% funding additionality, branch spending is about $720 per additional package, before allocating general organizational costs. Source

The model asks whether those contacts close a rescue gap beyond existing services. It assumes peers close 10% of the remaining gap with 90% effectiveness, reducing modeled overdose hazard by 9%. It values extra survival once over a finite ten-year horizon rather than awarding a new lifetime for each reported reversal.

With 25% of additional support allocated to peer work, our central estimate is about $172K per DALY in the Bay Area: one modeled equivalent year. The same additional paid hours support both response and workers’ household resources. We assume an $8.60 hourly resource gain after taxes, alternative pay, lost benefits and work costs, and a one-year paid-work window; negative resource changes remain possible. This improves the earlier health-only estimate modestly, rather than treating all payroll or reported job placements as extra welfare. More

What information has Code Tenderloin shared about its program?

Current program descriptions and the advertised $27 hourly ambassador wage support ongoing activity. Three original sponsor returns identify historical Code program spending, but not current standalone finances. Worker-hour logs, distinct people reached, emergency outcomes, paid tenure and net alternative resources would allow the estimate to be calibrated to actual additional activity. More

What is GiveBetter’s qualitative assessment of Code Tenderloin?

Trusted relationships, employment and reentry support can have value beyond the health pathways quantified here. The emergency-response mechanism deserves attention, but we would not infer cost-effectiveness from contact or reversal counts alone. Current funding and patient overlap with other providers are important next checks. More

1. What do they do?

Code Tenderloin provides employment preparation, training, casework, youth and reentry support, and community outreach. Its Community Ambassadors and Night Navigators assist people facing drug-related emergencies and seeking recovery services. Paid ambassador work also offers on-the-job training and career support.

Paid local outreach

The official page advertises about $27 an hour and case-manager and career-counselor support. It does not establish current worker counts, actual hours or tenure.

Additional timely response

The estimate concerns a selected high-risk population with a residual gap beyond existing public and partner response. Contacts, reversals and unique people are different units.

Net worker resources

Only wages above the alternative, after taxes, benefit withdrawal and work costs, receive household-resource credit. Training graduates and reported placements are not automatically extra jobs.

Scope of this review. The same paid peer hours support the health and worker-resource mechanisms; the gift is not spent twice. Other training, housing and long-run employment effects may matter but have no identified local causal increment in this calculation. Omitted effects are not proven zero.

2. Monitoring and information sharing

Current Code Tenderloin Inc, EIN 93-3030657

Current official program and IRS-derived identity. The organization describes paid ambassadors, emergency outreach, training and casework, and its current EIN agrees with the IRS-derived listing. The listing supplies no standalone original financial statements in this review.

Our assessment. This supports current activity, not a current expense series or funding offer. Missing index data does not prove inactivity, nonfiling or zero expense.

Center for New Music, a different legal entity

Three original self-filed sponsor returns. For years ended September 2023, 2024 and 2025, named Code program expenses are $3,957,514, $4,765,838 and $3,841,331. Sponsor total expenses are $4,424,251, $5,236,849 and $4,184,831.

Our assessment. These are historical sponsored-program context, not standalone Code finances or a current three-year Code expense mean. Part III program revenue excludes allocated contributions; it is not total project income.

Community Ambassadors and job-training clients

Advertised wages and uncontrolled program outcomes. The official ambassador page advertises about $27 per hour. Current organizational claims about job placement and retention have no causal comparison, verified current cohort denominator or measured pay increment here.

Our assessment. Income is modeled through explicit additional paid hours and net alternative resources. Reported placement, graduation and instruction counts are not automatically incremental jobs or durable earnings.

Other overdose-response settings

External naloxone decision model and post-overdose survivor cohort. Timely naloxone has a credible rescue mechanism. An external retrospective survivor cohort reports substantial later mortality, which argues against assuming every rescue restores an entire healthy lifetime.

Our assessment. These sources do not fit Code’s current cause-specific hazards, residual coverage gap or response effect. The finite competing-hazard model remains an explicit judgment; further observations should update it rather than create arbitrary new coefficients.

3. Qualitative assessment

Trusted local peers can assist during emergencies and connect people to services, while paid on-the-job training can improve workers’ resources relative to their alternatives.

Key reservations

  • Current standalone accounts, contract transfer and restrictions have not been reconciled.
  • Additional unique high-risk coverage and mortality effects are not locally identified.
  • Worker paid hours, tenure, alternative resources and benefit withdrawal are unmeasured.
  • Public and partner services can replace activity that a private donor imagines is additional.
  • Other training, housing and long-run earnings effects remain important unquantified channels, not proven zero.
  • Signed recipient burdens, displaced workers and shared rescue episodes require explicit follow-up.

Benefits not included in our estimate

  • Gross payroll or full advertised wages as net household gain
  • Uncontrolled job placements or later career pay as causal earnings
  • Separate worker mental-health gains overlapping resource welfare
  • A new lifetime for each recorded overdose reversal
  • Unmeasured housing, youth, training and caregiver effects
  • Unpriced partner and public resources as free societal inputs

4. What do you get for your dollar?

About $170K per DALY in the Bay Area

The current estimate retains the earlier explicit finite survival model because new accounts and employment information do not identify a new causal overdose effect. It adds a separately exposed worker-resource ledger rather than relabeling payroll as benefit. Both ledgers concern additional activity relative to existing funding and services. The illustrative donor-budget calculation is a reproducibility scale, not a recommended gift size or an accepted funding tranche.

DALY is our one-year comparison unit. QALY-based health gains and income-welfare equivalents retain their original assumptions; they are not measured clinical DALYs. These are uncertain estimates, not measured returns or verified donation offers.

How we calculate the estimate

Peer response and net paid-work resources: Funded peer hours = (100,000 × .25) / 60; additional hours = funded hours × .50. Health uses additional reached packages over a finite survival curve; income uses the same additional paid hours, net of alternative pay and offsets.
Additional paid hours = 208.3333; net health = 0.5642572509 all-US years; worker income = 0.0301216729 equivalent years; combined Bay = 0.5824913452 years. Ten donor-cost units divided by that Bay benefit gives $172K.

Model inputs and assumptions
Peer allocation and direct delivery
25% allocation; $60 per worker-hour (range: 10–50% allocation; old $45–$90 loaded-hour diagnostics). The advertised wage is about $27. Central payroll loading is 1.35 and direct nonwage delivery $23.55 per hour, giving $60. Six worker-hours buy an assumed offered person-year package. Institutional overhead and other activities stay in the remaining budget; no second overhead multiplier is added. Analyst judgment except the advertised wage.
Funding response and unique reach
50% additional hours; six hours per offered person-year (range: No additional activity; stronger public baseline). Central funded hours are 416.6667, yielding 69.4444 offered packages and 34.7222 additional packages. Repeated and unsuccessful contacts must fit inside the six-hour assumption. Funding response is distinct from a recipient’s existing rescue coverage. Analyst judgment except the advertised wage.
Finite health bridge
OD hazard .04; other-death hazard .03; residual gap .10; rescue .90 (range: One-year horizon; gap zero; high competing mortality). These are selected-cohort judgments, not estimates for all Code clients. The first service year reduces OD hazard by .0036. Both groups then return to the same hazards, including recurrent overdose. Utility .80, ten-year finite horizon, three-month service-start delay and 3% discount are explicit assumptions. Analyst judgment except the advertised wage.
Independent harms
0.00 health years per additional reached package (range: Response null with harms retained). Health gains are not new lifetimes per reported reversal. Signed harms remain when positive response fails. Applicant or worker burdens require additional gift-induced exposure, not unchanged baseline activity. Analyst judgment except the advertised wage.
Worker incidence and finite job window
1,000 paid hours per worker-year; one-year window (range: Shorter tenure; no net worker increment). 208.3333 additional hours correspond to .208333 worker-year equivalents at the assumed distribution. This is not an observed Code job duration or actual worker count. No long-run pay rise or placement probability is inferred from uncontrolled outcomes. Analyst judgment except the advertised wage.
Net hourly resource increment
$8.60 (range: Zero or negative net increment). Central $27 × .80 take-home − $10 alternative wage × .80 − $3 benefit withdrawal − $2 work costs. Only the $27 advertised wage is source-derived; other values are incidence judgments. Negative net resources must not be clipped. Analyst judgment except the advertised wage.
Annual household baseline and comparison
$25,000; 0.5 × log resource change (range: Alternative resource baseline; income-only and health-only). Each worker-year has $8,600 net resources: .5 × ln(1 + 8,600 / 25,000), discounted to the .75-year calendar midpoint. That midpoint includes the .25-year start lag once. Coefficient Giving’s comparison weight is normative, not clinical QALYs. Analyst judgment except the advertised wage.
Nested local benefit
98% Bay; 90% SF (range: Independent health and income shares; zero local attribution). Residence shares are judgments, not service addresses. Worker resources and reached-person survival can have different geography. SF belongs within Bay; reducing attribution never reduces the donor-cost numerator. Analyst judgment except the advertised wage.

Health, income and counterfactual diagnostics

  • Central: $172K Bay; $187K SF. Ten combined health and income-equivalent years, not measured clinical outcomes.
  • No net worker-resource increment: $181K Bay; $197K SF. Retains the earlier clinical central bridge and sets income to zero. Equal alternative resources or complete offsets can produce this case.
  • Funding replacement: No additional modeled benefit. No changed paid work, reached cohort or induced applicant burden. Donor cost is retained.
  • No residual rescue gap: Signed health loss remains possible. Positive peer-response health is absent while harms remain; worker resources are a separate channel, not attenuated by clinical success.
  • Negative net worker resources: Signed income loss. A stronger alternative job, taxes, lost benefits or work costs can make the resource increment negative. Net losses are retained in the combined denominator.

Uncertainty. The former health-only central estimate was about $181K Bay and $197K SF. The new $172K Bay price is modestly lower because a bounded paid-worker resource mechanism is included. It is not a new experimentally measured overdose effect or proof of better local outcomes. The original favorable $5K SF stress remains historical, not the current headline or a confidence bound. Zero or negative total benefit has no meaningful positive price.

Health and income effects

The shared funded-hours envelope is 100,000 × .25 / 60 = 416.6667 hours. The assumed additionality of .50 gives 208.3333 extra paid hours, or 34.7222 additional six-hour reached-person packages. Financial response is applied once, not once for survival and again inside its package effect. Both the service population and the workers can benefit from these same hours; this does not require allocating the gift twice.

For clinical benefit, baseline total hazard is .07 and service-year hazard .0664. After year one both arms face .07 again. Integrating the discounted survival difference over ten years gives 0.02 health years per additional package before .0005 shared harm. The resulting all-US net health is .5642572509 years. Averted repeat reversals do not each earn a fresh lifetime.

For worker resources, .208333 worker-year equivalents × .5 × ln(1 + 8,600 / 25,000) / 1.03^.75 gives 0.03 income-equivalent years. Actual allocated net incremental dollars are $1,791.67, not the entire advertised wage bill. The one-year job window and resource distribution are assumptions. Shorter tenure must change worker-equivalent counts and exposure consistently while preserving the same funded hours; it does not make resources recur after the job ends.

Workers and the reached cohort are distinct modeled recipients. No employee mental-health, financial-distress or job-placement welfare is added on top, and no future earnings of people whose survival improves are separately stacked. Extra applicant costs have their own signed gift-induced exposure; zero additional completed work alone cannot erase costs already incurred. All losses and benefits are assigned once to the same geographic and portfolio ledgers.

    5. Funding and previous grants

    The current independent recipient and services are identified, but historical sponsor accounts cannot substitute for the independent organization’s finances. Confirm current contracts, public support and extra peer-response capacity before using the assumed allocation of a gift as a recommendation.

    Reconcile current independent accounts and public contracts, then obtain a priced additional service and paid-work plan with recipient follow-up.

    No binding costed additional funding offer was located. A June 2025 CHOPP proposal names the former sponsor and mixes service units and components; a committee recommendation does not establish current transfer or actual spending. A February 2026 city announcement describes $500,000 of federal Community Project Funding for the workforce and wellness hub. This is public baseline support, not evidence of a remaining private gap. Current independent cash, restricted grants, contract execution and capacity require reconciliation. Complete societal costs and other program benefits remain incomplete.

    Historical sponsored-program spending

    The former sponsor’s original returns identify Code program expenses of $3,957,514 for the year ended September 2023, $4,765,838 for September 2024 and $3,841,331 for September 2025. These are not current standalone Code accounts or a comparable three-year operating-expense average. Sponsor Part III program revenue excludes allocated contributions and must not be read as total project income. The original returns are linked in Sources.

    This review does not establish a verified marginal funding offer or a complete history of grants.

    Donate

    Opens the organization’s giving page. A general donation may not fund the specific activity modeled here; confirm allocation with the recipient.

    6. Sources

    1. Official general giving and independent EIN. Code Tenderloin. Primary provider disclosure. Published: Undated; retrieved: September 8, 2026.
    2. California EO Business Master File: EIN 933030657. IRS. Primary regulator record; exact row preserved. Published: Current downloaded file; retrieved: September 8, 2026.
    3. Grantee partner updates and celebration 2025. SV2. Primary funder report. Published: June 2, 2025; retrieved: September 8, 2026.
    4. Community Ambassador Program. Code Tenderloin. Primary role and indicative wage disclosure. Published: Undated; retrieved: September 8, 2026.
    5. Treatment Navigation Guide. SFDPH. Current primary directory. Published: September 1, 2026 retrieved version; retrieved: September 8, 2026.
    6. Form 990, year ended September 30, 2024. Center for New Music / IRS. Primary filed return rendered by ProPublica. Published: Filed August 6, 2025; retrieved: September 8, 2026.
    7. CHOPP agreement request, contract 1000034290. SFDPH. Primary request, not expenditure record. Published: June 2, 2025; retrieved: September 8, 2026.
    8. Modeling overdose and naloxone distribution with fentanyl versus heroin. Coffin, Maya and Kahn. Primary model abstract. Published: 2022; retrieved: September 8, 2026.
    9. One-year mortality after ED treatment for nonfatal opioid overdose. Weiner et al.. Primary observational cohort abstract. Published: 2020; retrieved: September 8, 2026.
    10. Naloxone Distribution Project. DHCS. Primary state public-supply program. Published: Current page; retrieved: September 8, 2026.
    11. Overdose prevention and response,2025 totals. San Francisco AIDS Foundation. Primary partner baseline disclosure. Published: 2025 totals on current page; retrieved: September 8, 2026.
    12. 2025 impact report: Code Tenderloin partnership. San Francisco Community Health Center. Primary partner disclosure; not causal evaluation. Published: September 2025; retrieved: September 8, 2026.
    13. Original sponsor Form 990, year ended 2025-09-30. Center for New Music / IRS, mirrored by ProPublica. Self-filed sponsor return; not standalone Code accounts. Published: Filed 2026-08-13; retrieved: 2026-10-01.
    14. Original sponsor Form 990, year ended 2023-09-30. Center for New Music / IRS, mirrored by ProPublica. Self-filed sponsor return; not standalone Code accounts. Published: Filed 2024-08-14; retrieved: 2026-10-01.
    15. 2026 federal workforce and wellness hub funding announcement. San Francisco Recreation and Parks. Public funding announcement. Published: 2026-02-18; retrieved: 2026-10-01.
    16. Cost-effectiveness comparison methodology. Coefficient Giving. Normative comparison framework. Published: Current methodology; retrieved: 2026-10-01.

    Annual expenses: years and sources

    Average annual expenses (three consecutive fiscal years): Not available. Organization size is separate from the modeled cost-effectiveness of a donation.

    Code Tenderloin Inc.

    A standalone three-year expense series is not available. Historical fiscal-sponsor totals are excluded.