BeaconICT
Phone-first rules engine
6-question intake → rules-based fit engine → most appropriate available bed, ranked in seconds. Every handoff logged, every decline reasoned.
Kiosk + instant pass printing
Phone-less equity layer. Biometric enrollment prints a same-day paper pass — no City ID required, same fit engine behind it.
Try it live at beaconict.vercel.app · Same $500K the city is already budgeting — new math on what it buys · Toggle bottom-right to swap 09:00 baseline ↔ 13:00 incident.
Transition Bottleneck Dashboard
Where the system jams today — pulled from the live intake, shelter, and stabilization-docs APIs. Most of these jams are handoff failures in disguise: the bed exists but the gate blocks the referral, the referral succeeds but the decline is never logged, the decline is re-routed but the next step never closes. Each bottleneck below is named, sized, and paired with the BeaconICT response that closes its loop.
Beds vs. Reachable Beds (by provider)
Capacity exists. Access doesn't. Sobriety screens, referral rules, population restrictions, and clinical-assessment waits push 80% of nominal beds out of reach for the neighbors showing up today. Red numbers = reachable after gates.
Named Jam Points
The Response · Four Moves
One coordinated system, not four silver bullets. Every move is instrumented, measured, and visible to the whole coalition.
BeaconICT Coordination Platform
Phone-first public intake at beaconict.vercel.app — 6 questions, no account, case code + PIN for recovery. Rules-based fit engine ranks providers on eligibility, hours, barriers, capacity, and system stress. Closed-loop handoffs make every decline auditable. Native EN + ES.
Kiosk Access Layer
5 kiosks cover the phone-less: same intake, same fit engine, same case record — just a different front door. Biometric + paper-pass bridges the City ID gap. Without this layer, equity is gated by device ownership.
Scheduled Foster Protocol
Pre-enroll at DCF Independent Living at age 16. Milestones at 17, 18, 19, 20, 21. Blocker tasks auto-generate (ID, KanCare, BRIDGES); identity cliff neutralized before it happens; navigator relationship preserved through 21.
Coordinated Operating Model
Governance stays with United Way of the Plains. BeaconICT adds a shared intake record, a triage queue every partner can see, auditable decline reasons, and a weekly coalition review tied to this live dashboard.
Prioritization Model
The platform's fit engine ranks which Housing Partner is the best match for a given neighbor. This prioritization model handles the harder question: when reachable beds are fewer than neighbors waiting, which neighbor goes first, and why. Six transparent factors every partner can see and audit. AI surfaces the top candidates; a Navigation Partner case manager confirms every placement.
Six Factors · Weighted Score
Weights are defensible defaults — coalition reviews them quarterly. Any partner can see why a case ranks where it does. No black box.
Worked Examples
Safeguards
- Transparent weights. Published; reviewed quarterly by coalition; adjusted via governance process, not hidden tuning.
- Human confirms every placement. Model surfaces top-3 candidates per open slot. Case manager decides.
- Audit log. Every score computation stored for six months. Appeals process documented.
- Fairness monitoring. Weekly coalition review checks for demographic drift in rankings.
- No predictive profiling. Factors describe current situation, not predicted behavior.
Youth Handoff Failure Analysis
583 Kansas youth age out of foster care each year. The cliff is predictable. The failures are predictable. Nothing in the current system schedules for them. Today's storm disrupted foster placements for 4 aging-out youth — the exact moment the scheduled protocol is built to catch, before anyone falls through the cliff.
1 expected within 72 hrs
100% capacity
bridges issued
ahead of the disruption
case #2471· 17yo DCF exit · foster placement dropped · → BRIDGES waitlist top · pre-enrolled navigator dispatched · ETV claim pre-filledcase #2468· 19yo KanCare lapse · → flex-fund motel (3 nights) · re-enrollment task queued to DCF IL · decline reason logged on CrossRoads handoffcase #2465· 20yo cliff-proximate · CrossRoads full · → Family Promise overflow (warm handoff · youth-specific flag) · navigator contact reactivated
Foster Cliff by Age
Midwest Study national rate: 36% without stable housing by 26. Wichita's 42% by 21 outpaces it. Benchmark: extended foster care = 23% reduction.
FT Employment Gap
The employment ramp is real — but without housing stability, the youth can't sustain the job. The cliff isn't a labor market problem. It's an infrastructure problem.
Where the Handoff Breaks
The Scheduled Protocol (Ages 16 → 21)
| Age | Milestone | Dashboard action |
|---|---|---|
| 16 | DCF IL pre-enroll · navigator assigned | Profile created · kiosk biometric opt-in · 6-month check-in scheduled |
| 17 | Document prep · FT employment ramp start | City ID · SSA · birth-certificate queue verified complete |
| 18 | Exit from care · housing continuity plan locked | Kiosk final enrollment · BRIDGES / HumanKind / host-home slot tagged |
| 19 | Midpoint check · KanCare extension active | Employment verification · benefits continuity audit · adjust placement if needed |
| 20 | GED / postsecondary milestone · ETV claims submitted | ETV funding confirmed · housing stability score tracked |
| 21 | Graduation out of protocol · KanCare extension ends at 26 | Transition to adult pathway · navigator relationship preserved optionally |
Two Front Doors · One Case Record
Same intake, same fit engine, same closed-loop handoff queue. The access path adapts to the neighbor — the system doesn't care which door they used. Foster youth with phones use Door A; youth without data or a charged device use Door B at a kiosk.
A Phone Path · Primary Channel
B Kiosk Path · Phone-Less Equity Layer
Kiosk Access Layer · 5 Sites
The phone-first platform is the primary channel. These 5 kiosks are the equity backstop — for neighbors without phones, without data, or without a charged device. Biometric enrollment prints a same-day paper pass — no Wichita City ID required. Same intake, same fit engine, same case record. CrossRoads anchors the foster-youth coverage at 24/7.
2 sites inside the formal Navigation Partner network (Second Light, CrossRoads), 3 outside (City Hall, QuikTrip, ALL). QuikTrip + CrossRoads carry 24/7; the other three stack extended daytime hours. No single point of failure.
Use of Funds · $500K Pilot
$100K marketing is mandated; $400K is the operating envelope Oracle allocates. Every locked line maps to a measured KPI. The coordination platform itself is the Architect team's build — our dev line funds kiosk software + HMIS integration + kiosk↔platform adapter.
Dev line shown at $150K midpoint ($100K–$180K range). Reserve buffer absorbs swings. $100K marketing is mandated & flows through the Muse track.
Cost-of-Instability Analysis
Current emergency-service floor in Wichita (conservative Douglas County KS benchmark):
- $35K–$150K/yr per neighbor in transition (sustaining)
- $13K–$25K/yr per neighbor housed (Housing First)
- Net savings: $10K–$125K/yr per neighbor housed
- Neighbors in transition = 14.5–17.5% of Douglas County jail bed-days
Pilot KPI Framework · 90-Day Targets
| KPI | Baseline | 90-day target | Source |
|---|---|---|---|
| Median time-to-housing | 45 days | < 30 days | system-description · headline metric |
| Handoff completion rate | Untracked | > 80% | platform closed-loop data (new) |
| Reachable beds / total | 17 / 87 (20%) | 35 / 87 (40%) | shelter-inventory |
| 211 loop rate | ~33% | < 15% | intake-network operatorNotes |
| Platform intakes (phone-first) | 0 | 1,500 + | Architect platform |
| Kiosk enrollments (phone-less) | 0 | 300 + | pilot instrumentation |
| Median intake → bed match | Days (est.) | < 2 hrs | triage queue |
| Foster youth pre-enrolled (16–17) | 0 | 50 + | DCF IL roster |
| Emergency spend avoided (annualized) | — | > $500K run-rate | Douglas County benchmark |
Renewal Case · ROI Scenarios
The pilot funds itself. If year 1 hits KPIs, the cost-of-instability avoided in year 1 exceeds the $500K ask — year 2 is already paid for. Diversion % = share of $5.15M emergency-service floor deflected by the pilot. Foster pre-enrolled = 16–17 cohort onboarded at DCF IL intake during the 90-day window.
| Scenario | Diversion | Foster pre-enrolled | Year-1 ROI | Year-2 cumulative |
|---|---|---|---|---|
| Conservative | 10% | 50 | 1.0× | 2.1× |
| Base | 15% | 75 | 1.5× | 3.4× |
| Stretch | 20% | 100 | 2.0× | 4.8× |
If year 1 hits KPIs, the city has already paid for year 2. Renewal argument writes itself.
Renewal Trigger Conditions
Year-2 renewal is recommended if the 90-day pilot clears all three gates:
- Time-to-housing: median drops from 45 days toward the < 30-day target — handoff closure is working at the headline level.
- Handoff completion: > 80% of logged handoffs reach
completedorrerouted (with reason)— never silent declines. - Diversion economics: annualized emergency-spend avoided run-rate exceeds pilot cost — Conservative scenario or better.
If two of three gates clear, recommend a 60-day extension with targeted corrective actions. If one or zero, pilot sunsets honestly — the evidence package still leaves Wichita with handoff data it has never had.
Curveball Readiness · Distributed by Design
The 13:00 curveball hit exactly as forecast: severe weather event · ~60 displaced residents entering the shelter network · 211 call center overloaded · Open Door offline for emergency roof repair · Center of Hope absorbing overflow at reduced hours · CrossRoads near capacity as foster-placement disruptions push more youth into the system. The platform flipped currentMode = "incident", the fit engine re-ranked, and the dashboard's incident banner lit up. That's the state you're viewing by default — toggle "Show 09:00 Baseline" (bottom right) to see the pre-storm snapshot and back.
What happens automatically when the API flips to critical:
- Open Door offline (roof repair): fit engine drops Open Door from the ranked list; overflow routes to Union Rescue, Family Promise, and weather shelters surface automatically — no manual reconfiguration.
- Center of Hope at reduced hours: hours are a scored input on every recommendation; the engine ranks Center of Hope lower during the closed window and higher when it re-opens. Neighbors don't walk to a locked door.
- 211 overloaded: phone-first beaconict.vercel.app intake absorbs the spill — every minute of 211 wait time is a minute BeaconICT can instead be matching a case. No callback queue, no phone tree.
- CrossRoads near capacity · foster disruption in flight: prioritization model scores cliff-proximate youth highest; flex fund unlocks motel bridge vouchers; pre-enrolled 16–17 youth are already in the queue — this is Move 3 proving itself in real time.
- ~60 displaced (severe weather): surge hits the shared triage queue, not a single staff inbox. Exposure-weighted scoring floats unsheltered weather-exposed cases to the top; flex fund issues emergency motel vouchers; coalition review fires in < 60 minutes.
- 4 of 5 kiosks stay up: QuikTrip Downtown + CrossRoads retain 24/7 coverage; biometric portability means anyone enrolled at any kiosk authenticates at any other — no one becomes ID-less because one building closed.
The point: we predicted 4 of the 5 curveball elements in our design docs before the scenario dropped. The fifth — foster placement disruption — is the live validation of Move 3. Device-resilience on top of facility-resilience. Single points of failure are the problem we came to solve, not the thing we get caught by.
Why This Wins the Scoring Dimensions
| Dimension | How we score |
|---|---|
| Problem Understanding | Names handoff failure as the bottleneck, not capacity. Quotes Wichita's 45-day pace as proof the fix is specific. Cites every data point the challenge surfaces. |
| Solution Quality | Coordination platform + equity access layer. Phone or kiosk — either works. Prioritization rubric is transparent and auditable. Redundant by design. |
| Presentation & Polish | Air Capital palette. 6-tab navigable dashboard reading the same APIs the platform reads. Live on real API shape. One tab per required deliverable. |
| Adaptability | Native incident mode. Device-resilience on top of facility-resilience. Curveball response built into the architecture, not bolted on — click the toggle, see it fire. |
| AI Mastery | Rules-based fit engine · AI-generated plain-language recommendations · AI-assisted form filling · AI translation (EN/ES native) · AI-surfaced bottlenecks and prioritization · human confirms every placement. |