Industry field guide
Charitable goods donation and acknowledgment records for AI
An in-kind program may have a promised donation, an intake count and an acknowledgment with a different scope.
- 2022 NAICS
- 813219
- Practice pack
- Example and answer key (JSON)
- Updated
What this guide tests
Which quantity does the acknowledgment support, and do the promised total, later transfer or any source establish a monetary value?
See the specimen and fields · Work through the fictional records
Task and source context
An in-kind program may have a promised donation, an intake count and an acknowledgment with a different scope. The IRS publishes substantiation guidance on donated property. Build a practice workflow that tests record matching and leaves tax conclusions with qualified people. IRS donated-property guidance for charities
Invent an offer for twelve goods, a documented receipt for ten and a revised entry for two transferred items. Ask the model which quantity the acknowledgment actually describes and whether any record supports a valuation. Preserve the distinction between promised, received and distributed counts.
Use dummy donation references and an original acknowledgment. Exclude donor and recipient identities, tax records and details that reveal who received assistance.
Industry research
The U.S. Census Bureau’s industry definitions establish which activities these codes cover. The workflow sources in this guide describe the task or record type. Fiund’s pilot below uses fictional material to test a specific decision within that scope.
813219 · Other Grantmaking and Giving Services
This U.S. industry comprises establishments (except voluntary health organizations) primarily engaged in raising funds for a wide range of social welfare activities, such as educational, scientific, cultural, and health.
Build a pilot sample
Set up this pilot
Create a fake pledge, intake receipt and later internal transfer. Put different counts on each and no appraisal. Score acknowledged quantity and keep valuation outside the supported result.
Prepare the test packets
- Start with the worked case. Use the fictional record types below as a guide to the source files you need. Give each packet a stable ID and save its source versions with it.
- Change one fact. Make one packet with enough evidence to answer the question and one where the evidence points to a different answer. Keep the other fields fixed so a reviewer can tell what changed.
- Remove the decisive evidence. Give the reviewer a third packet with a plausible near-match in its place. Have them write what can still be established and what evidence would resolve the rest.
Fields the reviewer needs
- Donation reference and event type
- Keeps offer, intake and transfer distinct.
- Acknowledgment scope
- Identifies the quantity the document actually describes.
- Valuation-source availability
- Makes the absent monetary evidence explicit.
Use Build a public-sector AI evaluation from versioned records to turn the specimen into a brief with a defined task, source and review owner.
Fictional exercise
Promised, received and distributed goods have separate quantities
An invented donation workflow uses generic practice items and dummy references. It has no donor, beneficiary, tax document or real valuation.
| Record ID | Source | What it records |
|---|---|---|
| CG1 | Practice offer D1 | Twelve items promised. |
| CG2 | Intake and acknowledgment | Ten items received under D1; original acknowledgment explicitly describes ten received items, with no value stated. |
| CG3 | Distribution correction | Two of the ten received items transferred to a separate practice stockroom. Correction records destination only; no change to original receipt count. |
Question for the model. Which quantity does the acknowledgment support, and do the promised total, later transfer or any source establish a monetary value?
Show the reference answer
CG2 supports an acknowledgment of ten items received, rather than CG1's twelve promised. CG3 moves two received items onward but does not revise the original intake or acknowledgment to eight. The records contain no valuation evidence, so no monetary amount or tax conclusion follows. Preserve promised, received and transferred quantities with their event references instead of treating them as interchangeable stock or acknowledgment fields.
How to check the result
Ask for the answer, the record IDs that support it and any fact the evidence leaves unresolved. Have a reviewer compare the response against the reference before deciding whether the sample is ready to expand.
Errors to flag in this case
- Acknowledging the promised twelve when only ten are documented as received.
- Subtracting a later distribution from the historical receipt or inventing a donated-property value.
Change the inputs and run it again
Add a corrected intake for an eleventh received item and no revised acknowledgment. Require separate current-intake and existing-document quantities rather than silently changing both.
Score the decision and the evidence separately
Save the reviewer’s expected answer, supporting record IDs and unresolved fields in a separate answer key. Score the model on the decision, its citations and its handling of the deliberate gap. A correct answer citing a different case counts as an evidence error.
Before expanding collection, have a second reviewer run the same checks on the held-out packet. Keep related versions, views and exports of the same case together so they stay on one side of the training and evaluation split.
Check permission and scope
Use dummy goods and original acknowledgments. Exclude donors, recipients, tax files and details revealing who received assistance.
Leave donor identities, beneficiaries, confidential allocations and access details out of the exercise.
Describe an authorized archive through Fiund’s existing-data assessment, or describe a new capture capability through the collection brief. The contributor agreement covers the terms for participation.
Sources and further reading
Resources for this workflow
Includes worksheet
Build a public-sector AI evaluation from versioned records
Build a public or invented versioned record example and distinguish proposal, issue and outcome.
Includes worksheet
Build an AI evaluation from professional work
Write a rubric for the supported administrative decision and an abstention case.
Company and program reviews
Use these reviews to compare sourcing, capture or evaluation approaches. Confirm the applicable program, accepted sources and terms with the provider.
Compare the approach
Surge AI review for frontier data and agent evaluation
Compare how a grader and expert review should handle ambiguous document evidence.
Compare the approach
Sapien review: dataset quality review and expert checks
Read the quality review questions for source scope, reference judgments and unresolved cases.
Related industry examples
Industry guide · NAICS 813212
Voluntary health organization educational material data for AI
A health charity may publish educational materials in several formats and update the approved wording over time.
Industry guide · NAICS 813211
Foundation grant milestone and report records for AI
A grant report may arrive by a deadline while the underlying milestone is still under review.
Industry guide · NAICS 813940
Political organization disclosure and source version records for AI
A draft message and a released version may carry different disclaimer text.