Soldier DataAI Advisory Request a call

Know the ground before you move

AI planning, training and policy for public health, emergency management and community health teams.

We show you where AI belongs in the work you already do, train your staff, and write the rules. It starts with a fixed-price assessment.

AI Readiness Assessment$7,500 fixed fee · 3 weeksStarts with a free 30-minute call
Request the free 30-minute callor call (253) 948-8200

Read a sample report: health department emergency management hospital

James Soldier, Air Force and Air National Guard veteran, leads every engagement with associates from emergency management, public health, community health and nursing.

The AI practice is new. The planning experience is not. See the record ↓

SBA-certified SDVOSB · Lacey, Washington

01 — Today

Right now, nobody has the full picture.

Staff use several AI tools, and nobody tracks who uses what. Nobody is in charge of it.

02 — Inventory

First, we write it all down.

Every use of AI in your department, approved or not, drawn from staff interviews.

03 — Ranked

Then we rank it.

Each use is scored on payoff against risk. Drafting a plan section from sourced data scores well. Identifiable health data in a public tool does not.

04 — Plan

Then it becomes a plan.

Ninety days, five tracks, named owners and decision points, sized to fit alongside real work.

Inventory 30 of 40 candidate uses · illustrative, not client data

  1. 01Grant narrative first drafts
  2. 02Summarize public comment
  3. 03Synthesize listening sessions
  4. 04Mitigation plan sections
  5. 05Continuity plan refresh
  6. 06After-action report drafts
  7. 07Situation report drafts
  8. 08Translate health notices
  9. 09Surveillance summaries
  10. 10Outbreak FAQ first drafts
  11. 11Board packet summaries
  12. 12Secondary data profiles
  13. 13Resource line call sorting
  14. 14Clinic scheduling replies
  15. 15Shelter inventory checks
  16. 16Improvement plan drafts
  17. 17Exercise scenario drafts
  18. 18Plan crosswalk checks
  19. 19Policy gap analysis
  20. 20Public records redaction
  21. 21Meeting notes
  22. 22Grant report tables
  23. 23Site selection memos
  24. 24Community benefit reporting
  25. 25Hotline FAQ answers
  26. 26Case note summaries
  27. 27Eligibility screening
  28. 28Risk message drafts
  29. 29Partner outreach emails
  30. 30Legacy data migration

Ranked payoff against exposure · illustrative

Start here

Pays off. Low exposure, with a person checking.

  1. Grant narrative first drafts
  2. After-action report drafts
  3. Board packet summaries
  4. Exercise scenario drafts

With controls

Pays off. Needs a sign-off rule.

  1. Mitigation plan sections
  2. Synthesize listening sessions
  3. Surveillance summaries
  4. Public records redaction

When convenient

Safe. Small payoff.

  1. Meeting notes, not case meetings
  2. Partner outreach emails
  3. Grant report tables

Leave alone

The risk outweighs the payoff.

  1. Eligibility screening
  2. Case notes in a public tool
  3. Public alerts written by AI

90 days five tracks, named owners · illustrative

PolicyDirector or deputy

TrainingTraining lead

PilotProgram lead

ReviewCounsel or compliance

DecisionDirector

Week 1Decision · weeks 12–13

05 — Yours

Now describe your situation.

Pick your organization and up to three pressures.

1 — You are a

2 — What is pressing 0 / 3

A sketch · not an assessment · nothing leaves your browser

See result ↓

First moves, in order

  1. Put the health-data rule in writing within 30 days, and have your privacy officer review what was already entered.
  2. Draft plan sections with AI from sourced data, with a planner signing off on every figure. It shortens drafting, not the public process.
  3. Pilot AI synthesis of community input, checked against a hand-coded sample, and take the themes back to participants.
  4. Pilot AI synthesis of community input in an IT-approved tool, checked against a hand-coded sample, and take the themes back to participants.
  5. Build a grant-narrative workflow from your past applications, with a reviewer step before submission.
  6. Start with after-action drafts, not live reports. Any later situation-report pilot keeps your approval chain and a paper fallback.
  7. Write a plain-language policy against the NIST AI Risk Management Framework, and brief the board.
  8. Define who signs off on any AI-assisted public message before a tool is chosen.
  9. Turn activation and exercise logs into draft after-action reports and improvement plans, reviewed by the plan owner.
  10. Write requirements from your own use inventory before taking another demo.
  11. Favor configured tools over custom builds, and train two internal owners.
  12. Inventory every use of AI across your organization, approved or not.

90-day plan

Policy
Training
Pilot
Review
Decision
12345678910111213

Summary

Turn this into the real assessment →

Your choices carry over to the intake form. Nothing is sent until you send it.

AI at work

One task, shown end to end.

A grant funding notice goes in. A requirements matrix comes out. Every row cites its paragraph, and the model flags what it is unsure of.

In · funding notice excerpt

A.2Eligible applicants are state, tribal, territorial and local health departments and emergency management agencies, and nonprofit hospitals or community organizations applying with one.

B.1The project narrative shall not exceed 12 pages, 11-point font, single spaced.

B.2Applicants shall describe how the project serves people with access and functional needs, citing local data.

B.3Applicants shall include a work plan with quarterly milestones for the 24-month period of performance.

B.4A letter of commitment is required from each partner organization named in the narrative.

C.1A non-federal match of 25 percent is required. In-kind contributions are allowable.

D.1Applications are due by 5:00 p.m. Eastern time on the date in the funding notice. Late applications will not be reviewed.

E.1Applications will be scored on need (30 points), approach (40 points), capacity (20 points) and budget (10 points).

E.2The project narrative, not to exceed 15 pages, will be scored for a clear link between local data and proposed activities.

F.2Recipients shall not enter personally identifiable or protected health information into any system not approved by the program officer.

Out · requirements matrix Point at a row to see its source

  1. A.2

    Confirm we are an eligible applicant.

    Go or no-go check

    Eligible applicants are state, tribal, territorial and local health departments and emergency management agencies, and nonprofit hospitals or community organizations applying with one.

  2. B.1 · E.2

    Project narrative: 12 pages, 11-point, single spaced.

    Format check

    Flag B.1 says 12 pages. E.2 says 15. Ask the program officer. Write to 12 until answered.

    The project narrative shall not exceed 12 pages, 11-point font, single spaced.The project narrative, not to exceed 15 pages, will be scored for a clear link between local data and proposed activities.

  3. B.2

    Show service to people with access and functional needs, with local data.

    Narrative, need section

    Flag A person checks every figure against its source before it goes in.

    Applicants shall describe how the project serves people with access and functional needs, citing local data.

  4. B.3

    Work plan with quarterly milestones across 24 months.

    Narrative, approach section

    Applicants shall include a work plan with quarterly milestones for the 24-month period of performance.

  5. B.4

    One letter of commitment per named partner.

    Attachments

    Flag It does not say whether letters count against the page limit. Ask.

    A letter of commitment is required from each partner organization named in the narrative.

  6. C.1

    25 percent non-federal match. In-kind allowed.

    Budget

    A non-federal match of 25 percent is required. In-kind contributions are allowable.

  7. D.1

    Due 5:00 p.m. Eastern. No late applications.

    Submission checklist

    Flag The date is not in this excerpt. A person reads it off the notice.

    Applications are due by 5:00 p.m. Eastern time on the date in the funding notice. Late applications will not be reviewed.

  8. E.1

    Scored: need 30, approach 40, capacity 20, budget 10.

    Sets where the pages go

    Applications will be scored on need (30 points), approach (40 points), capacity (20 points) and budget (10 points).

  9. F.2

    No identifiable or health information in a system the program officer has not approved.

    Narrative, data section

    Flag This rule also governs any AI tool used on the project.

    Recipients shall not enter personally identifiable or protected health information into any system not approved by the program officer.

What the model was given

This excerpt and one instruction: list every requirement and scoring factor, cite the paragraph, flag conflicts. No client or health data. Only public text.

What a person checks

Every row against its paragraph. Every date, page limit and match figure. Each flag, before a question goes to the program officer.

What stays with people

The decision to apply. Every local figure. Partner commitments. Anything that identifies a person.

Worked example. The excerpt was written for this page and is not a real funding notice. An AI model drafted the matrix from it alone. On a real application, a person checks every row.

The record

Our team has written the plans you are writing.

155state facilities given continuity-of-operations plansRobert A. Soldier

$29Mfederal preparedness grant led as Principal InvestigatorFrances R.G. Limtiaco

400+responders trained and certifiedJames Soldier, Florida Air National Guard

75community health workers in a network built across seven countiesDaniel Hannawalt-Morales

Each figure is from that person’s earlier military, government or community health role, not from this firm’s work. Prior service and employers describe experience. They do not imply endorsement.

The people

James is your point of contact and brings in the associate whose field matches yours. About the firm

The AI practice is new. It has no AI clients yet, so the first engagement is small and fixed price.

Our AI work you can open today. VetsLearn, our free AI curriculum for veterans: 105 lessons, 16 working tools, 1,324 prompts. Every lesson teaches checking AI output before relying on it. Open it ↗

The assessment

AI Readiness Assessment

One department, office or program. The same for every field.

$7,500Fixed fee · 3 weeks · no follow-on obligation

  1. Before we start

    Scope

    • A written scope and fixed quote for your purchasing office
    • We sign your confidentiality agreement, or bring ours

  2. Week 1

    Listen

    • A one-hour kickoff. We ask who else must sign off, such as IT, counsel, records or compliance, and include them
    • Up to ten 45-minute interviews, by video or on site. In a small team that is everyone, plus your compliance, IT or finance contacts

  3. Week 2

    Review

    • A ranked inventory of the AI uses we find, approved or not
    • A data and rules review: what your rules cover and miss for health information, public records and other sensitive data, against the NIST AI Risk Management Framework

  4. Week 3

    Plan

    • Three recommended first projects and a 90-day plan
    • A one-page starting draft of a data rule, for your side to finish
    • A one-hour briefing for your director, board or commissioners

Your staff’s time45 minutes per person interviewed, plus the kickoff and the briefing.

If an emergency intervenesWe pause and resume when you are ready, at no charge.

Request the free 30-minute callSee a sample report ↓

For your purchasing officeThe fee is under the federal micro-purchase threshold ($15,000 since October 2025). Your own purchasing rules may be stricter. Vendor facts, scope and data handling on one page →

After the assessment

What usually comes next.

  • The full acceptable-use policy, through adoption, with your privacy officer, counsel, IT and records officer
  • It names your covered programs, business associate agreements, small-number suppression, confidential data outside HIPAA, and public records
  • Staff training, plus a Board of Health session
  • Community health assessment and improvement plan input, synthesized and checked against a hand-coded sample
  • A grant-narrative workflow from your past applications
Sample reportA county health departmentWritten for an invented organization. Open it →
  • After-action reports and improvement plans drafted from your activation and exercise logs
  • A continuity plan refresh, starting with the sections each department owes you
  • Hazard mitigation plan sections drafted from sourced data. The public process, state and FEMA review, and local adoption stay yours
  • A written rule keeping AI out of public alerts
  • Training for planners on checking AI output, and working when the tool is down
Sample reportA county emergency management officeWritten for an invented organization. Open it →
  • Community health needs assessment input, synthesized, checked against a hand-coded sample, and taken back to participants, who are told up front how their words will be used
  • First drafts of the implementation strategy and the community benefit report narrative that finance owns
  • Grant narrative and report workflows from your past applications
  • Work inside tools your IT and compliance teams have approved
  • Community health worker program uses that keep client records out of AI tools
Sample reportA hospital community benefit teamWritten for an invented organization. Open it →

Each project is scoped and quoted after the assessment.

The fine print

Questions buyers ask.

How do you handle what we tell you?
  • We do not record interviews. We take written notes.
  • The report names roles, not people, and quotes no one by name.
  • Your notes and documents never go into public AI tools. Any AI tool we use on them is a business account that does not train on your data, named to you before we start.
  • Notes and documents sit in encrypted storage open only to your engagement team.
  • We delete them 90 days after the final report, or sooner if you ask.
  • We sign your confidentiality agreement before the first interview, or bring ours.
  • Everything we deliver is yours.
Why not just use the free templates?

Use them. The Kansas Health Institute policy template, the GovAI Coalition templates and the NIST AI Risk Management Framework are good and free, and we build on them.

A template cannot tell you what your staff already do with AI, which uses are worth keeping, or who has to sign off. The assessment finds that out. Then the template gets filled in with facts.

Do you need our health information?

No. The assessment does not need protected health information. We work from interviews and written policies, and do not ask for case, client or patient records or examples from them.

If later work would touch protected health information, a business associate agreement and a system approved by your IT and compliance teams come first.

What does the work after the assessment cost?

It is not priced here. The assessment does not write your full policy or run a pilot. Those are projects it recommends, and we cannot price them honestly until it shows what each involves. Each then gets its own written scope and fixed quote. You can do them yourselves, hire someone else, or stop.

Can a grant pay for it?

Often, as a planning cost, but your grant terms decide. We give you a written scope and fixed quote that states the deliverables, so you or your state agency can check it against the grant.

The fee is under the federal micro-purchase threshold ($15,000 since October 2025), so federal grant rules generally do not require competing quotes. Your own purchasing rules may be stricter.

Do you work under other firms?

Yes. We subcontract to planning and consulting firms in these fields. W-9 on request.

How we think

Three things we tell every client.

Where the assessment sits in the NIST AI Risk Management Framework

Each part of the assessment answers one of the framework’s four functions.

Govern
Who owns the question, what rules exist, and what a policy must say.
Map
The inventory: every use, who does it, and who it affects.
Measure
How each recommended project would be measured against today’s work.
Manage
The ranking, the 90-day plan, the decision point, and the list of uses to leave alone.
What we tell clients not to build
  • A public chatbot, before your staff trust its answers.
  • A custom build, when a tool you own can do it.
  • Anything that decides about a person without a named human signing off.
  • Automation of a process nobody has written down.

Saying no early costs less than a pilot nobody uses.

Five questions before the next vendor demo
  • What data leaves our building, and where does it go?
  • Who checks the output, and how long does the check take?
  • What does it cost at our volume in year two?
  • If we leave, do we take our data and prompts with us?
  • Show it getting one wrong. What does that look like?

We resell no software and take no referral fees, so we can ask them with you.

Services

01

Strategy and readiness

Find out where AI pays off, and where it does not.

02

Workforce training

Teach your people to use AI and check its work.

03

Policy and governance

Write the rules before the tools spread.

04

Custom builds

Build the tool when off-the-shelf does not fit.

05

Ongoing advisory

Keep a standing adviser as the tools change.

Start with a free 30-minute call.

Request the free 30-minute call(253) 948-8200

ServicesThe assessmentPurchasingAboutContactPrivacy

Soldier Data LLC · SBA-certified SDVOSB · UEI EGQKWX55SE84 · CAGE 21TK6 · Lacey, Washington · (253) 948-8200 · [email protected]. No reseller ties, no referral fees. Our public AI work: VetsLearn.