What changed
FACT (from the provided award record): DHS obligated $2,885,874,392.17 to the STATE OF FLORIDA DIVISION OF EMERGENCY MANAGEMENT to reimburse state, local, tribal and territorial governments and certain private non-profits for pandemic emergency protective measures — emergency medical care, medical sheltering, vaccine administration and distribution, and community engagement. INFERENCE: Florida DEM is the recipient/pass-through; the actual filers are hundreds of counties, cities, school boards, special districts, tribes and PNP hospitals/shelters that must substantiate their own cost lines to draw and keep that money.
Why now
HYPOTHESIS (not established by the source text): COVID Category B costs were incurred in 2020-2022, so the live work in 2026 is not initial filing but closeout, obligation adjustments, appeals, and the DHS OIG audit tail — which historically runs many years after obligation. The urgency is defensive: money already received can be de-obligated. INFERENCE: this makes the COVID-specific slice a wasting asset, and the durable business is the same engine applied to every future declared disaster (Categories A-G), which Florida generates almost annually.
Converging signals
Three things meet at one point: (1) a very large pass-through award with a defined subrecipient class (FACT, award description); (2) a state portal layer (FloridaPA.org) sitting on top of FEMA Grants Portal, meaning filers work in a state-specific system with state forms and state deadlines (INFERENCE — not in the source text); (3) an entrenched consultant class that already bills to assemble these packages (INFERENCE from industry structure, not from the provided data). Note: the `demand_evidence` array's high-similarity items are Medicaid entitlement grants and unrelated disaster-repair awards. They are semantic neighbours, not evidence for THIS opportunity. I am scoring demand from the one on-point FORCED BUYER item only, and saying so.
Customer pain
HYPOTHESIS: a county finance office that spent $40M on medical sheltering and vaccine PODs has that spend scattered across a payroll system, a purchasing system, and a pile of contracts. FEMA eligibility rules are per-cost-line (was the labor overtime or straight time? was the employee budgeted? was the contract competitively procured? was the vaccine work within the incident period?). Reconstructing this by hand years later, for an auditor, is the pain. No complaint threads or job postings were supplied to corroborate this; treat it as unproven pain until the founder confirms it in customer calls.
Who pays
Three candidate buyers, in descending order of reachability: (a) FEMA PA consulting firms who want to cut their own labor cost per engagement — reachable, buy fast, no procurement; (b) PNP hospitals, shelters and small municipalities without in-house grant staff — reachable, and their purchase sits under formal bid thresholds; (c) county/school-board grant-compliance leads — biggest wallet, slowest path. The founder should NOT anchor on (c).
Solved today
INFERENCE (industry structure, not sourced from the input): disaster-recovery consultancies — Tidal Basin, Hagerty, Witt O'Brien's, ICF, CDR Maguire — assemble these packages, frequently billing a percentage of the award or an hourly rate that is itself reimbursable to the subrecipient as a management/administrative cost. Document handling happens in FEMA Grants Portal and the state portal, plus spreadsheets and shared drives.
Why current solutions are bad
HYPOTHESIS: because the consultant's fee is often charged back to the grant, the subrecipient perceives the consultant as free — which is the single hardest fact for this business. It is not that the current solution is bad; it is that it is expensive to the federal taxpayer and invisible to the buyer. The genuine defect is that eligibility errors are discovered years later, at audit, when de-obligation is the remedy and the consultant has moved on. Nobody currently sells a pre-flight check that runs BEFORE submission.
Proposed product
A hosted tool that takes payroll registers, general-ledger/AP exports, contract PDFs and timesheets; normalises them into a cost-line ledger; classifies each line against FEMA PA policy (category, incident period, force-account labor vs. contract, budgeted vs. unbudgeted, procurement method); scores each line's audit exposure; and generates the submission binder plus a per-line eligibility narrative with source-document citations. The pre-flight flag list — 'these 340 lines will be disallowed, here is why' — is the product's actual value, not the document assembly.
MVP version
Narrow it hard: Category B force-account labor and overtime only, one payroll export format (Tyler Munis or ADP), one state (Florida). Ingest CSV, apply the labor-eligibility ruleset, output (1) a flagged-exceptions report with dollar exposure and (2) a formatted labor-cost workbook matching what FDEM/FEMA reviewers expect. No portal write-integration in v1 — read and produce, do not submit.
30-day build
Do not write the classifier yet. Get the ground truth first: pull published DHS OIG audit reports of COVID PA subrecipients and code every disallowance into a rule (this is public-records work, the founder's strength). In parallel, run 15-20 discovery calls — 10 with FEMA PA consultants, 10 with Florida county grant managers and PNP hospital finance staff. The kill question to ask verbatim: 'who pays for your PA package today, and does that cost come out of the grant?' If every answer is 'the grant pays,' stop here.
60-day build
Build the labor-eligibility engine against the OIG-derived ruleset. Validate it by running it against 2-3 real, already-audited claim packages obtained from a friendly consultant or via public records request — the engine must independently rediscover the disallowances the OIG actually found. That reproduction is the demo and the entire sales argument. Anything less and this is a spreadsheet with opinions.
90-day revenue plan
Sell the pre-flight as a fixed-fee audit-exposure assessment, not as software: '$7,500, you send exports, I return a flagged-exception report with dollar exposure per line.' Service first, product underneath. Two or three of these fund the build and generate the case study. Convert the consultants into resellers/white-label users at a per-engagement licence once the reproduction demo lands. Software subscription revenue is a 6-12 month story, not a 90-day one — do not pretend otherwise.
Distribution path
Direct outbound to named FDEM subrecipients (the applicant list is public), FEMA/FDEM applicant briefings and PA conferences, the Florida Association of Counties and Florida Government Finance Officers Association, and — most importantly — subcontracting to the consultancies rather than fighting them. Publishing an 'OIG disallowance taxonomy' derived from public audit reports is a credible, zero-cost content wedge that reaches every grant manager in the country.
Pricing hypothesis
Assessment engagement $5,000-$15,000 depending on claim size. Software licence $12,000-$25,000/year per subrecipient, or per-engagement licence $1,500-$3,000 to consultants. Avoid contingency/percentage-of-recovery pricing: it invites state procurement scrutiny and drags the founder into the consultant business model he is trying to undercut.
Technical difficulty
Moderate on the code, high on the domain. Parsing payroll/GL exports and building a rules engine is a few weeks of work. Encoding FEMA PA eligibility correctly enough that an auditor's finding is reproduced is the hard part, and it cannot be shortcut with an LLM alone — every rule must trace to policy text or an OIG finding. Document handling of contracts and timesheets is OCR-plus-extraction work that is now tractable.
Legal / regulatory risk
Real and specific. Producing an 'audit-ready' eligibility opinion edges toward the territory of accountants and grant-compliance professionals. Mitigate by positioning as a documentation and exception-flagging tool that the subrecipient's own staff or auditor signs off on, with explicit non-opinion language. Ingesting payroll data means PII — standard handling, not a blocker. Do not touch percentage-of-recovery fees on federal funds without counsel.
Platform dependency
Low. FEMA Grants Portal and FloridaPA.org are government systems with no platform owner who can deplatform the tool. V1 does not integrate with them at all. If a portal API is later used, the dependency is on a federal system, not a commercial gatekeeper.
Founder fit
Strong on shape, strong on background. The founder has already read a federal mandate, identified a forced filer class, and built a portal-submission business (FMCSA ELDT). Public-records mining of OIG audit reports is squarely in his wheelhouse, as is the fire/emergency-services vocabulary that PA work is written in — he can hold a credible conversation with a county emergency manager on day one. The mismatch is the sales motion: he sells through demonstrated value, and this buyer class buys through relationships and procurement. The consultant channel is the bridge.
Breakout potential
The COVID Category B slice is a declining asset. The engine is not. FEMA PA runs on every declared disaster, in every state, forever, with the same cost-line eligibility logic across Categories A-G; every state has a FloridaPA.org equivalent. Prove the ruleset on one hurricane in one state and the same product sells 49 more times without a rewrite. That, and not the $2.9B headline, is the reason to build this.
Final recommendation
CONDITIONAL BUILD, with the framing changed. Do not build a 'COVID Category B audit-defense tool' — that is a shrinking market entered late. Build a FEMA Public Assistance cost-line eligibility engine, use the COVID closeout and OIG audit tail as the entry wedge and case-study generator, and sell first to consultants and PNP filers rather than to counties. The whole thesis rests on one unvalidated fact: whether subrecipients experience the consultant as free. Spend 30 days and roughly zero dollars answering that before writing the classifier. If consultants pay for the tool, this is an excellent fit for the founder's proven government-mandate playbook and a 50-state replication story. If the only buyer is a county procurement office, kill it.
Next action
Pull every DHS OIG audit report of a COVID-19 Public Assistance subrecipient, code the disallowances into a structured rule table, and use that table as both the product's ground truth and the opening artifact in 20 discovery calls — 10 to FEMA PA consultancies, 10 to Florida county grant managers and PNP hospital finance leads. Ask each one who pays for their claim package today and whether that cost is charged to the grant.