craftloop
After DutyFor veterans, since October 2025

From a box of military records to a ranked claim plan.

Veterans preparing a VA disability claim face a stack of records and a rating schedule written for lawyers. Most of the work is reading: finding the diagnosis, the in-service event and the medical link for each condition, then working out what is still missing. After Duty does that reading, so a veteran understands their own information, whether they prepare the claim alone, with a veterans service officer or with an accredited attorney.

Visit afterduty.app

After Duty is an educational resource. It isn't legal advice, doesn't file claims or connect to VA, and doesn't replace a VSO or accredited attorney. It's an independent app, not affiliated with or endorsed by the U.S. Department of Veterans Affairs.

Designed and built by Craftloop

After Duty on the web and iPhone for the synthetic veteran D. Griff: the home screen with an estimated combined rating, monthly pay and next steps, and the phone showing PTSD's three-part evidence check.

The problem

A strong VA claim needs three things for every condition: a current diagnosis, an event in service, and a medical link between them. The proof is buried across hundreds of pages of records.

When a claim goes in missing information, the VA has to ask for more records or exams, and every extra round of review adds time before benefits arrive. Knowing what the claim needs before it's filed helps shorten that wait.

What we built

One codebase, three platforms
A Next.js app ships as the website and, through Capacitor, as the iOS and Android apps.
AI where it reads, code where it decides
Gemini extracts facts from each document. Claude finds conditions, estimates ratings and powers a chat that cites its sources. Presumptive rules, rating limits, VA's combined-rating math and the change log are plain, tested code, so the numbers a veteran sees are reproducible.
Value-ranked next steps
Each evidence gap is tested the way a VA rater would test it, then ranked by how far it can move the estimate.
Incremental by design
Content fingerprints mean only new or changed documents are re-read, which keeps re-analysis fast and cheap.
A knowledge base that corrects itself
When expert review disproves something the AI said, the correction is enforced in code on every later analysis.
Honest by default
Every estimate carries its confidence, says when a key test is missing, and reminds the veteran that only VA decides.

How a record becomes a plan

When a veteran adds a document, the analysis runs in this order. AI does the reading; plain, tested code does the deciding.

ExtractGemini reads the document and lists the facts, each tagged with its source
IdentifyClaude finds the conditions the evidence supports and merges duplicates
EstimateClaude estimates each rating against the rating schedule
Apply the rulesPresumptive rules, rating limits and VA's combined-rating math
Find gapsLists the evidence each condition is missing
Review and rankTests each gap the way a VA rater would, then ranks what's left
AI stepPlain, tested code
Deterministic: presumptive rules, rating limits, VA math and timeline diffs Incremental: content fingerprints skip documents that haven't changed Traceable: facts carry their source document, and answers cite 38 CFR

The screens

Scroll sideways. Every screen shows a synthetic veteran, D. Griff.

After Duty home screen showing a 90% estimated combined rating, $2,362 estimated monthly pay, claim readiness at 50%, and the highest-value next step.
See where your VA claim stands
Next Steps screen ranking evidence gaps; the top step is a missing nexus letter for sleep apnea that could support a 50% rating.
Spend your time on the step worth the most
Conditions list with PTSD at 70%, Asthma at 60% with a PACT Act badge, and Chronic Migraines at 50%, each showing diagnosis, in-service and nexus checks.
Every condition, checked against VA's three-part test
PTSD detail screen: VASRD code 9411, 70% estimated rating, 95% confidence, and strong diagnosis, in-service event and nexus evidence.
Know why each estimate holds up
Scenarios screen comparing a 90% rating and $2,362 a month for filing three ready conditions with 100% and $3,939 a month after closing two high-priority gaps, a $1,577 difference.
See what closing your gaps is worth
Ask AI conversation explaining what sleep apnea needs for a 50% rating, citing the veteran's sleep study and 38 CFR 4.97.
Ask about your claim. Get answers with sources.
Documents screen with options to upload files or take a photo, and a box to write a personal statement that becomes evidence.
Bring every record into one place
Claim timeline listing analysis updates by date, including PTSD rating estimate 50% to 70% and evidence gaps closed.
Nothing in your analysis changes silently
Learn hub listing five free guides: how VA rates disabilities, Intent to File, preparing for the C&P exam, free VSO help, and gathering records.
Plain-English VA guides, free for every veteran

How it grew

  1. Started, to make sense of one veteran's records
  2. Rebuilt for the web, iPhone and Android
  3. Renamed After Duty, on afterduty.app
  4. Store and site screens rendered from the real app, with a synthetic veteran

Built with

Designed and built by Craftloop, every layer: the web and mobile apps, the cloud services and the AI.

  • Next.js
  • React
  • Capacitor
  • Spring Boot on Cloud Run
  • Cloud SQL
  • Firebase passwordless auth
  • Claude on Google Vertex AI
  • Gemini
  • Stripe
  • RevenueCat

About 1,900 automated tests.