One prompt, four days, 3,000 counties: 79 assumable VA homes matched
BryanX (Twitter) · 2026-09-22
Read original1 prompt to @Muse 4 days ago. 'Find every Veteran Home Loan in US and match it with current homes for sale.' It's gone to 3000 counties public records site, scraped every home for VA flags, then matches it with homes for sale.
Business model
From a single instruction, Muse spent 4 days scraping public records across 3,000 counties for VA loan flags and matched them against 1,275 homes for sale, surfacing 79 assumable low-rate matches with per-home savings estimates; the author's planned realtor outreach had not run yet (author-reported).
The agent runs a multi-day public-records sweep and cross-matches listings into a qualified lead list with per-item savings math; outreach stays a separate, human-approved step.
Workflow
- Define the target property type, data scope and match conditions in one instruction.
- The agent sweeps public records over days and cross-matches active listings with per-home math.
- Spot-check accuracy before deciding on outreach.
Adaptation plan
No sale or revenue is reported; the list is the asset. Record-site rules and match accuracy need spot-checks. AgentFlow has not independently deployed this adaptation.
Starter prompt
Written by AgentFlow; not the original author’s prompt. Not independently deployed in Muse or Grok.
Screen public records for properties matching my criteria (attached), then cross-match with homes currently for sale. For each match give the key data and the savings estimate, compiled into a list. Run a small area first for my spot-check; scale up only if accuracy passes.
Running costs and pitfalls
No sale or revenue is reported; the list is the asset. Record-site rules and match accuracy need spot-checks. AgentFlow has not independently deployed this adaptation. No independent cost benchmark is available. Check subscriptions, permissions and external actions before running a workflow.