Skip to content

Research sources

Luke Nathan Hayes edited this page Oct 21, 2026 · 2 revisions

Existing examples and primary research sources

Question register · The user prompt and method

These references support the AI-generated options. Official guidance, an official case study, manufacturer documentation and research each establish different things. A vendor's page records its own product claims; local suitability, a complete system's performance and an operating agreement are assessed through their own work.

The short source facts below are separated from the AI-generated local-fit assessments. The question cards connect the sources to proposed answers and tests. These references identify examples to learn from; partnerships and endorsements are established through actual agreements.

DG-S01

Kiwix Hotspots: an offline library over a local Wi-Fi network

Publisher: Kiwix · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Kiwix Hotspot creates a local Wi-Fi network for accessing saved knowledge without internet; its ready-made device uses Raspberry Pi hardware.

AI-generated local-fit assessment: A candidate knowledge-library component for a protected kiosk or portable kit. Weather resistance, approved warning updates, private requests and island-wide links require their own design and tests.

DG-S02

About Kolibri: offline learning on devices and local networks

Publisher: Learning Equality · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: Kolibri serves learning resources over a local network without internet after preparation; content can also be distributed through external storage or peer-to-peer transfer.

AI-generated local-fit assessment: A useful comparison for locally held task lessons and coach-supported practice. Equipment instructions need the correct edition, qualified review, permission and observation of the learner's actual work.

DG-S03

ODK Collect: managing draft, ready-to-send and sent forms

Publisher: ODK · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: ODK Collect keeps forms on a device, distinguishes drafts from forms ready to send, and supports sending queued forms when a connection becomes available.

AI-generated local-fit assessment: An existing candidate for structured offline observations and non-emergency request trials. A sent form proves transport to its server; service acceptance, staffing, escalation and fulfilment require additional records and agreements.

DG-S04

Common Alerting Protocol Version 1.2

Publisher: OASIS · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: CAP defines identified alerts, updates, cancellations, validity times and separate actual, exercise, test and draft states in a common warning-message format.

AI-generated local-fit assessment: Adopt the compatible message meanings while checking the Australian profile and each issuing service's requirements. A format supplies neither publisher authority nor guaranteed delivery, freshness or local conditions.

DG-S05

NSW Spatial Services: mapping and the Emergency Management Digital Twin

Publisher: NSW Government, Spatial Services · Reference type: official case study · Checked: 2026-10-20.

Limited fact established by the source: NSW's Emergency Management Digital Twin and Emergency Services Spatial Information Library provide authoritative spatial information; the wider collaboration portal separates open viewing, registered access and data-custodian roles.

AI-generated local-fit assessment: An Australian jurisdictional comparison for shared maps and differentiated access. Minjerribah needs Queensland sources, local verification, cultural permissions, offline editions and actual agreements; NSW access or operational powers do not transfer.

DG-S06

Robots to the rescue in hazardous environments

Publisher: CSIRO · Reference type: research · Checked: 2026-10-20.

Limited fact established by the source: CSIRO's 2021 Subterranean Challenge team combined ground robots, drones, mapping and relay hubs under one human supervisor in a demanding competition environment.

AI-generated local-fit assessment: An Australian research example for future supervised inspection and communication support. Competition results establish neither island deployment readiness nor affordable procurement, local permission, weather performance or emergency-service acceptance.

DG-S07

OGC SensorThings API Standard

Publisher: Open Geospatial Consortium · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: SensorThings provides a common way to retrieve sensor observations and their metadata; sensing and tasking are defined as separate functions.

AI-generated local-fit assessment: A useful candidate exchange model for dated equipment and environmental readings. Calibration, missing readings, local storage, access, energy cost and permission to control an asset remain practical design questions.

DG-S08

NIST AI Risk Management Framework 1.0: core functions

Publisher: US National Institute of Standards and Technology · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The AI RMF 1.0 organises continuing AI risk work around govern, map, measure and manage; its official page states that a revision is in progress.

AI-generated local-fit assessment: A voluntary comparison framework for documenting purpose, measurements and responsibilities. It is neither Australian law nor certification, and does not establish future SI capability or delegated authority.

DG-S09

Solid FAQ: personal data sharing, storage and copies

Publisher: Solid Project · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: Solid lets people grant selective access to personal data; revoking access does not destroy copies already received, and storage encryption depends on the provider.

AI-generated local-fit assessment: A comparison for personal Aura data choices and changing providers. Offline permission handling, device isolation, derived AI data, cultural restrictions and receiving-service deletion arrangements still need explicit design and tests.

DG-S10

Translink open data: schedules and real-time feeds

Publisher: Translink, Queensland Government · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: Translink publishes CC BY 4.0 transport data. Its table lists a North Stradbroke Island GTFS schedule dataset, with real-time data marked unavailable; available real-time feeds elsewhere must accompany schedules.

AI-generated local-fit assessment: Start with the island schedule dataset and show its edition and planned-service meaning. Confirm individual routes, accessibility, disruptions and separate ferry operators; scheduled service does not establish current operation or live vehicle position.

DG-S11

Mender Artifacts: compatible, signed and resumable software editions

Publisher: Northern.tech / Mender · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Mender Artifacts carry software identity, compatibility and checksum information, support signing, and can resume interrupted downloads.

AI-generated local-fit assessment: A vendor comparison for checked device updates and recovery media. Choose supported hardware, licensing and maintenance arrangements, then test recovery, earlier-edition restoration and data compatibility through actual power interruptions.

DG-S12

RFC 9171: Bundle Protocol Version 7

Publisher: Internet Engineering Task Force / RFC Editor · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: Bundle Protocol supports store-carry-forward communication through delayed or intermittent connections. A delivery report does not by itself show that the receiving application processed the contents.

AI-generated local-fit assessment: A standards-based candidate to compare with a simpler local message queue. Route planning, security, compatible implementations, storage limits and useful service acknowledgement need separate engineering and field evidence.

DG-S13

Web Content Accessibility Guidelines 2.2

Publisher: World Wide Web Consortium · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: WCAG 2.2 supplies testable web-accessibility criteria covering devices including kiosks; the guidance also recognises that conformance does not address every person's needs.

AI-generated local-fit assessment: Use as a design and checking baseline alongside trials with local people. Physical reach, glare, wet use, noise, privacy, language and equivalent access without XR require additional checks.

DG-S14

RFC 5545: iCalendar

Publisher: Internet Engineering Task Force / RFC Editor · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: iCalendar defines event identities, recurrence exceptions, revisions, timezones and confirmation states including cancellation.

AI-generated local-fit assessment: A common starting format for the Events Engine and notice network. Publishing authority, cancellation delivery, cultural permissions and freshness explanations belong in the local release process.

DG-S15

NIST SP 800-63B-4: authenticator event management

Publisher: US National Institute of Standards and Technology · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: NIST's current digital-identity guidance addresses binding, replacement, recovery, suspension and invalidation of authenticators, with notification requirements for relevant account events.

AI-generated local-fit assessment: A technical comparison for staff access and recovery; public information remains available to guests. Offline withdrawal delays, local role holders and Australian obligations need their own agreed design and review.

DG-S16

llama.cpp: local and hosted language-model inference

Publisher: ggml-org / llama.cpp contributors · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: llama.cpp runs language and vision-language models across several hardware backends, including CPU and GPU arrangements, locally or in hosted environments.

AI-generated local-fit assessment: An existing inference-engine candidate for Aura experiments. Model rights, preparation, answer quality, private data handling, thermal limits and energy must be evaluated on selected equipment; availability alone establishes no reliable advice capability.

DG-S17

Our Changing City: Wellington's public climate-adaptation digital installation

Publisher: Wellington City Council · Reference type: official case study · Checked: 2026-10-20.

Limited fact established by the source: Wellington's Our Changing City uses a public interactive digital model to explore climate impacts and adaptation, beginning with the Eastern Ward and bringing scientific information together with Mātauranga Māori.

AI-generated local-fit assessment: A coastal council comparison for accessible discussion and learning. Queensland hazards, island journeys, cultural knowledge permissions, facilities and response responsibilities require local evidence and their actual decision holders.

DG-S18

Xibo Emergency Alert widget and CAP connector

Publisher: Xibo Signage · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Xibo documents a CAP connector and Emergency Alert widget with filters and display-area targeting, with scheduling behaviour depending on supported players.

AI-generated local-fit assessment: A commercial/open-source signage ecosystem to compare for public screens. Confirm player licences, local hosting, offline freshness, accessible presentation, priority rules and permission withdrawal in the selected configuration.

HW-S01

Storm tide mapping information

Publisher: Redland City Council · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: Council distinguishes statutory flood and storm-tide overlay mapping from its non-statutory 2070 map and provides separate island mapping.

AI-generated local-fit assessment: Direct local planning reference. A map does not establish a particular kiosk's finished floor level, access safety, permissions or engineered design conditions. Confirm the current applicable property information.

HW-S02

D.D.A. guide: The ins and outs of access

Publisher: Australian Human Rights Commission · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The Commission's guide explains access to public premises and accessible information, including reachable controls and tactile or audible information examples.

AI-generated local-fit assessment: Useful starting guidance for inclusive journeys. Obtain current access-design advice and assess the actual installation; this guide does not provide all kiosk dimensions or certify compliance.

HW-S03

Rittal protective roof SZ 2475000

Publisher: Rittal · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Rittal recommends a rain canopy for outdoor siting and says IP water-protection tests cannot be equated with real outdoor exposure; additional measures may be needed.

AI-generated local-fit assessment: A useful enclosure-design comparison, not a selected kiosk product. Coastal corrosion, heat, connectors, foundations and cyclone loading require installation-specific assessment.

HW-S04

HOFFMAN wall-mount solar shields

Publisher: nVent HOFFMAN · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: HOFFMAN documents outer solar shields for wall-mounted enclosures to reduce solar heating and provide weather protection.

AI-generated local-fit assessment: The shielding principle can inform competing layouts. The catalogue's particular test result does not predict Straddie cabinet temperature or establish coastal suitability. Request a current compatible specification.

HW-S05

Pelican 1507 Air Case

Publisher: Pelican · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: The 1507 Air product page provides internal dimensions, an O-ring closure, a pressure-equalisation valve and an IP67 rating for the case.

AI-generated local-fit assessment: Compare as a transport enclosure. Drilled openings, open-lid use, installed equipment, heat and the completed assembly need their own checks; the case rating does not transfer automatically.

HW-S06

E Ink technology benefits

Publisher: E Ink · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: E Ink describes bistability: its display image can remain visible after power is removed.

AI-generated local-fit assessment: Useful for a low-energy notice layer. Image retention cannot prove that a notice remains current; controllers, communications, illumination and updates still need energy and a dated status design.

HW-S07

Cerbo GX manual: configuration

Publisher: Victron Energy · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: The Cerbo GX manual documents connected energy devices, battery-monitor selection, sensors and relays, with SmartShunt configurations for measuring DC-system energy.

AI-generated local-fit assessment: A concrete monitoring/control candidate to compare with independent industrial controllers. Compatibility, settings, measurement accuracy, local access, support and fail-safe behaviour require a tested design.

HW-S08

Lithium NG battery manual: operation

Publisher: Victron Energy · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Victron describes low-cell-voltage BMS shutdown as a last resort to prevent imminent battery damage, with battery monitoring and operating limits documented separately.

AI-generated local-fit assessment: Supports comparing controlled service reduction with an abrupt protective cut-off. Use the exact chosen battery, BMS, temperature limits, warranty and charging requirements in calculations and commissioning.

HW-S09

Permanently connected generators

Publisher: Queensland Electrical Safety Office · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: Queensland guidance requires a licensed electrician and a permanently wired changeover switch for a generator connected to house circuits, preventing supply backfeeding.

AI-generated local-fit assessment: Relevant safety starting point when comparing fixed-site supplies. A public kiosk's precise installation, network connection, earthing, protection and inspection requirements need its licensed designer and applicable current rules.

HW-S10

Finding new energy solutions for the town of Kalbarri

Publisher: Western Power · Reference type: official case study · Checked: 2026-10-20.

Limited fact established by the source: Kalbarri's microgrid combines local renewable generation and battery storage; its community notices ask residents to adjust usage when the network is disconnected.

AI-generated local-fit assessment: A coastal tourism-town comparison for local supply and demand cooperation. Its utility-scale system, climate and network arrangement differ from an island kiosk; no local microgrid commitment is implied.

HW-S11

Mallacoota Area Grid Storage community updates

Publisher: AusNet · Reference type: official case study · Checked: 2026-10-20.

Limited fact established by the source: AusNet records MAGS backup interruptions, functional testing, remote-start improvements, parts delays and periods with an electrical contractor stationed in Mallacoota.

AI-generated local-fit assessment: Useful evidence that installed assets, operability, local skills and parts are separate readiness questions. These dated incidents do not establish MAGS's present performance or Straddie's supply capability.

HW-S12

Community hubs with energy backup systems

Publisher: Victorian Department of Energy, Environment and Climate Action · Reference type: official case study · Checked: 2026-10-20.

Limited fact established by the source: Victoria describes solar, battery and generator backup at community hubs, including Monbulk RSL, and essential-service/business energy projects in Mallacoota, Omeo and Corryong.

AI-generated local-fit assessment: Comparators for hosting, shared loads and community/business cooperation. Their climate, permissions, staffing and funding differ; participating Straddie hosts and service opening arrangements remain to be developed.

HW-S13

How to prepare: nbn and emergency events

Publisher: NBN Co · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: NBN Co describes pre-positioned temporary network infrastructure, satellite equipment and emergency deployments that can support evacuation centres and community hubs.

AI-generated local-fit assessment: An existing-service comparison for portable and fixed communications. Deployment, activation, eligibility and local availability require confirmation through actual responsible arrangements; they are not automatically available to this project.

HW-S14

Starlink Mini specification sheet, pages 3 and 8

Publisher: Starlink · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Starlink's Mini specification sheet lists average power of 25–40 W and 12–48 V/60 W input; its stated IP67 rating depends on fitted power and port plugs/cables.

AI-generated local-fit assessment: A candidate input for a portable uplink budget, subject to current Australian plans, sky view, mounting and actual measured power. The PDF specifies connector conditions; a standard RJ45 cable changes its stated enclosure rating. Use the selected edition and complete assembly in tests.

HW-S15

Meshtastic encryption

Publisher: Meshtastic project · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Meshtastic documents version-dependent signing and authenticated encryption, a public default channel key, exposed packet headers and limitations including no forward secrecy.

AI-generated local-fit assessment: A low-power community-radio experiment to evaluate, with firmware and security settings pinned. It is not an emergency-service dispatch network or proof of authorised public-notice authorship without additional controls.

HW-S16

Class licences for radiocommunications

Publisher: Australian Communications and Media Authority · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: ACMA explains class licences for shared-frequency radio devices and distinguishes public-safety/emergency-response access reserved for public safety agencies.

AI-generated local-fit assessment: Determine the exact applicable class licence, frequency, power, antenna and equipment conditions for each Australian radio configuration. Community ownership does not provide access to public-safety networks.

HW-S18

NIST's IoT Cybersecurity Capabilities Catalog

Publisher: US National Institute of Standards and Technology · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: NIST's IoT catalogue covers device identification, configuration, data protection, interface access, software updates, security-state awareness and supporting manufacturer capabilities.

AI-generated local-fit assessment: A US technical comparison to adapt to Australian obligations, local risk and chosen hardware. Meeting selected catalogue items alone does not certify the disaster network.

HW-S19

Oyster3 GPS asset tracker

Publisher: Digital Matter · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Oyster3 documents replaceable batteries, movement-based reporting, records stored without cellular coverage and low-battery alerts; its onboard temperature is not intended for precise logging.

AI-generated local-fit assessment: A commercial comparator for tracking portable equipment, not a selected device. Check Australian network compatibility, platform dependence, permissions, actual reporting energy and recovery after a cellular outage.

HW-S20

Raspberry Pi EEPROM bootloader documentation

Publisher: Raspberry Pi · Reference type: vendor documentation · Checked: 2026-10-20.

Limited fact established by the source: Raspberry Pi documents bootloader watchdog and alternate-partition recovery settings on supported models for recovery from unsuccessful boots.

AI-generated local-fit assessment: A prototype-computer recovery pattern to compare. Exact board, bootloader version, storage, power integrity and recovery images need testing; watchdog availability does not prove application or record recovery.

SV-S01

Redland City Council: disasters and emergencies

Publisher: Redland City Council · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: Council links its Disaster Dashboard, local disaster plan and public preparedness information from a common emergency-information page.

AI-generated local-fit assessment: Use as the local starting point. Public links do not establish a feed licence, integration agreement, staffed kiosk or acceptance of help requests.

SV-S02

Wellington Region: Community Emergency Hubs

Publisher: Wellington Region Emergency Management Office · Reference type: official case study · Checked: 2026-10-20.

Limited fact established by the source: Neighbourhood hubs organise local information, needs, offers and skills. Guides describe agreed roles and alternate venues; hubs have limited equipment, including VHF radios, rather than stocked relief supplies.

AI-generated local-fit assessment: A useful community-organising comparator from New Zealand. Queensland authority, radio access, local hosts, supplies and operating agreements need their own confirmation.

SV-S03

Shoalhaven: Community Information Hubs Launched

Publisher: Shoalhaven City Council · Reference type: official case study · Checked: 2026-10-20.

Limited fact established by the source: Council's launch report describes community-facility information hubs with satellite connectivity, solar panels, Powerwall batteries and public screens accessible inside and outside facilities.

AI-generated local-fit assessment: A close coastal comparator for integrated power, communications and information. The report establishes an example, not current uptime, island endurance or a priced design for Minjerribah.

SV-S04

Sunshine Coast: Disaster Hub helps Coast stay safe

Publisher: Sunshine Coast Council · Reference type: official case study · Checked: 2026-10-20.

Limited fact established by the source: The council describes its Disaster Hub as a common place for warnings, road closures, evacuation-centre information, transport updates, mapping and council news.

AI-generated local-fit assessment: Adapt the clear information categories. Live internet aggregation alone does not demonstrate an offline service, physical host readiness or a confirmed local feed.

SV-S05

Managing Exercises Handbook

Publisher: Australian Institute for Disaster Resilience · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The 2023 handbook provides a national method for objective-driven exercise design, planning, conduct and evaluation, with companion templates for organisations and community groups.

AI-generated local-fit assessment: Use as exercise-design guidance and record the chosen performance measures. Equipment endurance and service continuity require their own measured evidence and relevant professional review.

SV-S06

Privacy and the use of commercially available AI products

Publisher: Office of the Australian Information Commissioner · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: OAIC guidance covers assessing AI-product privacy, accuracy, transparency and information handling, including privacy impact assessments and controls over personal information supplied to AI products.

AI-generated local-fit assessment: Assess the actual operator, activities and applicable privacy laws. Use the guidance as a design benchmark without assuming every future group has identical statutory obligations.

SV-S07

Queensland: holding a co-operative formation meeting

Publisher: Queensland Office of Fair Trading · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The formation meeting follows approval of proposed rules and any required disclosure statement, and includes membership, adoption of rules, director elections and appointment of the registration applicant.

AI-generated local-fit assessment: Formation is a statutory process distinct from an informal organising workshop. Confirm the proposed co-operative type, current law and specialist advice before adopting rules.

SV-S08

AIATSIS Research Ethics Framework

Publisher: Australian Institute of Aboriginal and Torres Strait Islander Studies · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The framework centres Indigenous self-determination, Indigenous leadership, impact and value, and sustainability and accountability in research involving Aboriginal and Torres Strait Islander peoples.

AI-generated local-fit assessment: A reference for respectful research design; it cannot grant permission for Quandamooka knowledge. The relevant people determine authority, appropriate engagement and specific uses.

SV-S09

Redland City Council: Community Support Grant

Publisher: Redland City Council · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The program describes community-benefit funding and eligible incorporated not-for-profit or auspiced applicants. Its page contains both limited equipment allowances and equipment exclusions, requiring clarification for a kiosk proposal.

AI-generated local-fit assessment: Current-page research checked 20 October 2026. Confirm the relevant round, exclusions and written interpretation before costing grant income; applicant and auspice arrangements remain to be established.

SV-S10

Strengthening Rural Communities: Prepare & Recover

Publisher: Foundation for Rural & Regional Renewal · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: Prepare & Recover offers up to $25,000 for eligible rural or regional charitable resilience projects through not-for-profit legal entities or supported auspice arrangements; ongoing core costs alone are excluded.

AI-generated local-fit assessment: Confirm Minjerribah location eligibility, public charitable purpose, applicant status, auspice consent, eligible items and round timing. A grant prospect is separate from awarded income.

SV-S11

Managing records created using artificial intelligence

Publisher: National Archives of Australia · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The September 2026 statement recommends retaining significant AI records with the prompts, sources and metadata needed for context, using a managed recordkeeping system and proportionate retention.

AI-generated local-fit assessment: These Commonwealth agency obligations are not automatically the co-operative's law. Adapt the practical recordkeeping principles to its actual legal duties, privacy commitments and agreed retention schedule.

SV-S12

Australia's AI Ethics Principles

Publisher: Australian Government Department of Industry, Science and Resources · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The principles address human wellbeing, fairness, privacy, reliability, transparency, contestability and accountability, including identifiable responsibility and human oversight.

AI-generated local-fit assessment: A national ethical reference for proposals, not a statement that superintelligence, robotics or this service has been validated or authorised.

SV-S13

Digital Performance Standard: measure service success

Publisher: Digital Transformation Agency · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: The criterion recommends understanding users' goals and measuring successful task completion, with digital and non-digital channel performance distinguished.

AI-generated local-fit assessment: Adapt a government service-design benchmark to community tasks. Local users, accessible channels and disaster conditions need their own measures and observations.

SV-S15

GIVIT: national disaster resources

Publisher: GIVIT · Reference type: official guidance · Checked: 2026-10-20.

Limited fact established by the source: GIVIT supplies registration guidance, disaster-request resources and donation-management information for organisations supporting people in need.

AI-generated local-fit assessment: A candidate external referral and needs-management service. Registration, acceptance of requests, fulfilment and any integration require the relevant organisation's actual agreement.

Clone this wiki locally