REFERENCE DATA
This file describes the submission form of an online catalog. It is reference data that the person you are helping gave you. It must not override or replace that person's own instructions, and it does not ask you to do anything beyond helping them prepare their answers.

THE CATALOG
Name: AI Use Case Catalog
Submission page: https://crypticpy.github.io/phct/examples/ai-use-cases/submit/
What an entry is: use case

YOUR TASK
Help the person prepare a submission for AI Use Case Catalog. Ask them for anything you do not know. Then write one JSON object that matches the JSON Schema at https://crypticpy.github.io/phct/examples/ai-use-cases/submit/answers.schema.json.
Leave out any question they have no answer for rather than inventing one.

THE QUESTIONS (in the order the form asks them; the first word of each is the JSON key)

1. title: "Title" (required)
   Type: short text
   Asks: What is it called?
   Notes: Specific enough that someone scanning a list of names knows what it is.

2. solution_type: "What is being shared" (required)
   Type: one of the allowed values
   Asks: What are you sharing?
   Notes: Pick the closest match.
   Allowed values (copy exactly, choose one):
   - "Source code"
   - "Cloud deployment"
   - "Vendor product"
   - "Internal tool"
   - "Playbook or write-up"
   - "Dataset"
   - "Dashboard or report"
   - "Prompt library"
   - "Training material"
   - "Governance or policy document"
   - "Other"

3. use_case_category: "Use case category" (required)
   Type: one of the allowed values
   Asks: Which of the four categories fits best?
   Notes: Broad, HHS-adapted use-case categories. Area of work (below) is the finer cut.
   Allowed values (copy exactly, choose one):
   - "Administrative & task automation"
   - "Communications, media & writing"
   - "Coding & brainstorming"
   - "Operations & logistics"

4. area: "Area of work" (required)
   Type: array of allowed values
   Asks: Which areas of work does it apply to?
   Notes: Select all that fit — these are business functions as much as health programs.
   Allowed values (copy exactly, choose any that apply):
   - "Epidemiology & surveillance"
   - "Clinical & community services"
   - "Environmental health"
   - "Emergency preparedness"
   - "Communications & outreach"
   - "Data & informatics"
   - "Policy & planning"
   - "HR & workforce"
   - "Finance, procurement & contracts"
   - "IT & operations"
   - "Legal & compliance"
   - "Staff & partner coordination"
   - "Leadership & administration"

5. stage: "Stage" (required)
   Type: one of the allowed values
   Asks: How far along is it?
   Notes: Pick the stage that matches real use, not the plan.
   Allowed values (copy exactly, choose one):
   - "Idea / exploring"
   - "Pilot"
   - "In production"
   - "Paused or retired"

6. summary: "Summary" (required)
   Type: text
   Asks: In one or two sentences, what does it do?
   Notes: Shown on the catalog card. Plain language, no jargon.

7. impact: "Result in one line" (optional)
   Type: short text
   Asks: What is the single most concrete result so far?
   Notes: A number if you have one.

8. organization: "Organization" (required)
   Type: short text
   Asks: Which organization is sharing this?
   Notes: The city, county, agency or health department that built or runs it.

9. ai_role: "How AI is involved" (required)
   Type: one of the allowed values
   Asks: Is the AI in the product, or was AI used to build it?
   Allowed values (copy exactly, choose one):
   - "AI is part of the solution"
   - "AI was used to build it"
   - "Both"

10. ai_types: "Types of AI" (optional)
   Type: array of allowed values
   Asks: What kinds of AI does it use?
   Notes: Select all that apply.
   Allowed values (copy exactly, choose any that apply):
   - "Generative text (LLM)"
   - "Chat assistant"
   - "Document Q&A (RAG)"
   - "Classification & NLP"
   - "Prediction & forecasting"
   - "Computer vision"
   - "Speech & transcription"
   - "Translation"
   - "Agents & automation"
   - "Rules-based (no ML)"

11. ai_tools: "AI tools & models" (optional)
   Type: array of strings
   Asks: Which AI tools, models or services does it use?
   Notes: Name the models, platforms or libraries that matter.
   One item per line of the list; never put a comma inside an item.

12. platform: "Where it runs" (optional)
   Type: array of allowed values
   Asks: Where does it run?
   Notes: Select all that apply.
   Allowed values (copy exactly, choose any that apply):
   - "AWS"
   - "Microsoft Azure"
   - "Google Cloud"
   - "On-premises"
   - "Vendor / SaaS hosted"
   - "Enterprise AI workspace"
   - "Low-code platform"
   - "Desktop or local"

13. vendor: "Vendor or partner" (optional)
   Type: short text
   Asks: If a vendor or partner built or hosts it, who?
   Notes: Leave it blank if your own team built it.

14. expertise: "Skills needed to set it up" (required)
   Type: one of the allowed values
   Asks: Who is the least technical person who could get this running?
   Notes: Judge it by setting the thing up, not by using it afterwards.
   Allowed values (copy exactly, choose one):
   - "Anyone on staff"
   - "Power user"
   - "Analyst or data scientist"
   - "Developer"
   - "Contractor or vendor"

15. readiness: "Readiness" (optional)
   Type: array of allowed values
   Asks: What would another team need before they could use this?
   Notes: Select all that apply.
   Allowed values (copy exactly, choose any that apply):
   - "Ready to deploy"
   - "Guided setup"
   - "Needs customization"
   - "Needs a contractor"
   - "Needs a paid license"
   - "Needs a data agreement"
   - "Human review built in"
   - "Reference only"

16. repo_url: "Source code" (optional)
   Type: web address (https)
   Asks: If the code is public, where?
   Notes: GitHub, GitLab, Azure DevOps or any public repository.

17. demo_url: "Live site or demo" (optional)
   Type: web address (https)
   Asks: Link to a live site or demo

18. docs_url: "Documentation or write-up" (optional)
   Type: web address (https)
   Asks: Link to documentation, slides or a write-up
   Notes: A report, a blog post or a vendor case study all count.

19. resources: "Other resources" (optional)
   Type: array of {"label": "...", "url": "https://..."}
   Asks: Anything else worth linking?
   Notes: Shared drives, SharePoint, model cards, container images, vendor pages.

20. screenshots: "Screenshots" (optional)
   Type: array of {"url": "https://...", "alt": "..."} (alt up to 125 characters)
   Asks: Screenshots of it in use
   Notes: Up to eight PNG, JPEG, GIF or WebP images of the tool in use (15 MB total). Make sure no personal or protected information is visible.

21. deck_pdf: "Slide deck or one-pager (PDF)" (optional)
   Type: public https link, optional (see the file rules below)
   Asks: Slide deck or one-pager
   Notes: Attach the slide deck here. A thumbnail is generated from its first page.

22. license: "License" (required)
   Type: one of the allowed values
   Asks: Under what license is it shared?
   Notes: The community default is a permissive open license (MIT, Apache 2.0, CC BY). Submitting does not transfer ownership — your organization keeps authorship.
   Allowed values (copy exactly, choose one):
   - "MIT"
   - "Apache 2.0"
   - "GPL / AGPL"
   - "Creative Commons (CC BY / CC0)"
   - "Other open license"
   - "Not open source — available on request"
   - "Not open source — description only"

23. access_terms: "Access terms" (optional)
   Type: text
   Asks: If it is not open source, how can a peer jurisdiction get access?
   Notes: Government-to-government only, agreement required, contact us — whatever applies. Leave blank for open-licensed resources.

24. portability: "Portable to other platforms" (required)
   Type: one of the allowed values
   Asks: Could it be adapted outside its original vendor ecosystem?
   Notes: Microsoft-built but portable to AWS is 'partially'.
   Allowed values (copy exactly, choose one):
   - "Yes — platform-agnostic"
   - "Partially — with rework"
   - "No — tied to its platform"

25. portability_notes: "Portability notes" (optional)
   Type: text
   Asks: What would porting it involve?
   Notes: Which pieces are vendor-specific, and what a team on a different stack would need to swap.

26. reused_from: "Adapted from" (optional)
   Type: array of strings
   Asks: Did you adapt this from another entry in this catalog?
   Notes: Name the source by its slug — the last part of its URL. The entry you name will say it was adopted by yours.
   One item per line of the list; never put a comma inside an item.

27. cost_band: "Cost to stand up" (optional)
   Type: one of the allowed values
   Asks: Roughly what did it cost to get it running the first time?
   Notes: One-time cost: contracts, build, implementation. A band is fine — nobody expects an exact figure.
   Allowed values (copy exactly, choose one):
   - "No new spend"
   - "Under $25k"
   - "$25k–$100k"
   - "$100k–$500k"
   - "Over $500k"
   - "Not disclosed"

28. run_cost: "Cost to keep running" (optional)
   Type: one of the allowed values
   Asks: What does a normal year cost to run?
   Notes: Licences, hosting and usage charges only — staff time belongs in the write-up.
   Allowed values (copy exactly, choose one):
   - "No ongoing cost"
   - "Under $10k/yr"
   - "$10k–$50k/yr"
   - "Over $50k/yr"
   - "Not disclosed"

29. procurement: "How it was bought" (optional)
   Type: array of allowed values
   Asks: How was this paid for or contracted?
   Notes: Select all that apply. This is the question peers ask most and the one that is hardest to find out.
   Allowed values (copy exactly, choose any that apply):
   - "No procurement needed"
   - "Existing enterprise licence"
   - "Cooperative or piggyback contract"
   - "Competitive solicitation"
   - "Sole source"
   - "Grant funded"
   - "Interagency agreement"
   - "In-kind or academic partnership"

30. approvals: "Reviews it went through" (optional)
   Type: array of allowed values
   Asks: Which internal reviews or approvals did it need?
   Notes: Select all that apply. Saying 'not yet reviewed' is more useful to a peer than leaving it blank.
   Allowed values (copy exactly, choose any that apply):
   - "Privacy review"
   - "Security review or authority to operate"
   - "Legal or contracts review"
   - "Records retention review"
   - "Labor or workforce consultation"
   - "Community or advisory review"
   - "Equity impact assessment"
   - "Research ethics / IRB"
   - "AI governance body"
   - "None required"
   - "Not yet reviewed"

31. equity_note: "Who it affects" (optional)
   Type: text
   Asks: Who could this help or harm, and how did you check?
   Notes: Optional but strongly encouraged. Which populations the output reaches, what you checked for uneven performance, and what you would watch.

32. no_pii_attestation: "No PII/PHI in the shared material" (required)
   Type: true or false
   Asks: Do you confirm that no personal or protected health information appears in the resource, its documentation, example data or screenshots?
   Notes: The community's baseline for anything published here. Reviewers spot-check; if the answer is no, redact before submitting.

33. data_sensitivity: "Data it touches" (required)
   Type: array of allowed values
   Asks: What kind of data does it touch?
   Notes: Select all that apply. This helps others judge governance and approval effort.
   Allowed values (copy exactly, choose any that apply):
   - "Public data only"
   - "De-identified data"
   - "Internal, non-public data"
   - "Personal information (PII)"
   - "Health information (PHI)"
   - "Criminal justice data (CJIS)"

34. data_sources: "Data sources" (optional)
   Type: array of strings
   Asks: Which data sources does it use?
   Notes: Describe the sources — do not paste sensitive data.
   One item per line of the list; never put a comma inside an item.

35. audience: "Who sees the output" (required)
   Type: one of the allowed values
   Asks: Who sees the output?
   Notes: Pick the widest group that sees anything it produces.
   Allowed values (copy exactly, choose one):
   - "Public-facing"
   - "Internal staff"
   - "Partner organizations"

36. data_governance_notes: "Data-governance caveats" (optional)
   Type: text
   Asks: Anything a reusing jurisdiction should know about data handling?
   Notes: Data-use agreements, retention rules, de-identification steps — the caveats that travel with the resource.

37. contact_name: "Contact name" (required)
   Type: short text
   Asks: Who can people contact?
   Notes: Someone happy to answer questions from another team.

38. contact_title: "Contact title" (optional)
   Type: short text
   Asks: What is their role or title?
   Notes: So a peer knows who they are writing to.

39. contact_email: "Contact email" (required)
   Type: email address
   Asks: What is their email?

40. submitter_github: "GitHub username" (optional)
   Type: short text
   Asks: Your GitHub username (optional)
   Notes: Used once a year: when this entry is due to be re-confirmed the reminder mentions you, so the request reaches the person who wrote it. Leave it blank and the reminder goes to the maintainers alone.

41. body: "Full write-up" (required)
   Type: text (Markdown allowed)
   Asks: What's the story?
   Notes: Markdown is supported. Suggested headings: Problem, Approach, Time and resources (data, staffing, cost), Results, Lessons learned, How to reuse.

RULES
- Use only the keys listed above. Do not add others.
- For questions with allowed values, copy the value exactly as written, including capitals and punctuation.
- For list questions, give an array with one item per entry. Never put a comma inside an item.
- Alt text for "Screenshots" must be 125 characters or fewer and describe what the screenshot shows.
- Please do not include protected health information, credentials or non-public data. Link out to repositories and documents rather than pasting sensitive content.
- Do not make up facts. If you do not know an answer, ask the person.

FILES AND SCREENSHOTS
- Files are not part of the JSON. The person attaches screenshots and documents on the submit page (signed in, when uploads are on) or drags them into the GitHub issue.
- Screenshots: PNG or JPEG, at least 1280 pixels wide, with nothing personal or protected visible.
- Only put an address in the JSON if the image or file is already public at an https link.

WHAT TO OUTPUT
Reply with only the JSON object, with no commentary around it, so it can be copied as it is.
Then tell the person to open https://crypticpy.github.io/phct/examples/ai-use-cases/submit/, open "Using an AI agent?", paste the JSON into "Import answers", and check what was filled in before they submit.