What changed
FACT (from provided source text): DHS has an open assistance award ASST_NON_4399DRFLP00000001_070 of $2,254,896,254.51 to the STATE OF FLORIDA DIVISION OF EMERGENCY MANAGEMENT, described as 'GRANT TO LOCAL GOVERNMENT FOR REPAIR OR REPLACEMENT OF DISASTER DAMAGED FACILITIES.' FACT: sibling awards of the same description exist at far larger scale — $35.30B (ASST_NON_4339DRPRP00000001_070, Puerto Rico GAR), $21.99B (US Virgin Islands), $17.37B (New York DHSES). HYPOTHESIS/INFERENCE: these are FEMA Public Assistance (PA) grant obligations under the Stafford Act, where the state is the recipient/grantee and the money is passed through to local subrecipients. The pass-through structure — not any new rule — is what creates the filer class.
Why now
INFERENCE: nothing in the source text states a deadline, and I will not invent one. What is FACT is that ~$2.25B is already obligated to one state agency and tens of billions to others, and obligated federal money cannot be drawn down until subrecipients file. The paperwork exists the moment the award exists. HYPOTHESIS: PA obligations of this size are multi-year and continuously replenished by each new declared disaster, so the filer pipeline is recurring rather than a one-shot deadline scramble. The 'why now' is therefore weaker than a rulemaking with a compliance date — this is a standing, always-on burden, not a cliff. Score speed_to_revenue on the standing burden, not on urgency the text does not support.
Converging signals
Three things meet at one point: (1) a very large obligated federal award to a state agency [FACT: usaspending ASST_NON_4399DRFLP00000001_070]; (2) a defined class of parties who must file to touch it — Florida counties, municipalities, special/water-control districts, school boards and private non-profits, as subrecipients of FDEM [INFERENCE from award description + standard PA structure]; (3) a state-operated submission portal layered on top of the federal one [INFERENCE: FEMA Grants Portal is federal and documented; FDEM's DEMES/FloridaPA.org subrecipient portal is asserted in the convergence description and NOT verified in any provided source — treat as unverified until the founder loads the site himself]. The same triangle repeats in PR, USVI and NY per the sibling awards, which is the strongest structural signal in this input: the shape is not Florida-specific.
Customer pain
HYPOTHESIS (no complaint threads or job postings in the provided evidence — demand_evidence contains only FUNDED MANDATE and FORCED BUYER award records, plus one off-topic HHS anti-kickback RFI): the pain is that PA reimbursement is documentation-gated. A subrecipient submits a Request for Public Assistance, then Project Worksheets, then must substantiate every dollar with invoices, force-account labor timesheets, equipment logs, and — the killer — proof that procurement followed 2 CFR 200.317-327. Missing or non-compliant procurement documentation triggers Requests for Information, delays drawdown by months, and at closeout or OIG audit can be deobligated, meaning the local government must repay money it already spent. I have NOT been given a single complaint, forum post, or job ad substantiating this. Per the scoring rules, a compelled filer class scores demand high on the mandate alone — but I am labelling the *mechanism* of pain as inference, not fact.
Who pays
Three candidate buyers, in descending order of realism. (1) BEST: the disaster-recovery consulting firms and grant-management shops that already do this work by hand — they bill hourly or on percentage and their margin improves directly if software eats the document-assembly hours. They buy software, they buy fast, and there is no procurement office. (2) The subrecipient itself — a county emergency manager or finance director. Reachable, but see kill arguments: their consultant is often free to them. (3) The state agency (FDEM). Reject as primary channel — that is exactly the government procurement office the founder profile rules out. The honest read is that the paying customer is the *consultant*, and the subrecipient is the end user. That is a real business but it is a different business than 'per-filing fee from 500 Florida local governments.'
Solved today
HYPOTHESIS (drawn from domain knowledge, not from the provided sources): subrecipients either (a) hire a disaster-grant consultant — Hagerty, Tidal Basin, Witt O'Brien's, IEM, AC Disaster Consulting — who staffs the recovery with contract grant managers; (b) assign it to existing finance/EM staff who work out of Excel, SharePoint and the FEMA Grants Portal UI; or (c) buy horizontal grant-management SaaS (AmpliFund, Euna/eCivis) that tracks awards generically but does not know what a Project Worksheet needs. None of these three is verified by a source in this input.
Why current solutions are bad
HYPOTHESIS: horizontal grant SaaS is award-agnostic — it will remind you a report is due but cannot tell you that your emergency debris-removal contract lacks the cost-or-price analysis that 2 CFR 200.324 requires, which is the specific finding that gets money clawed back. Consultants know that, but they are expensive, thin on the ground immediately post-disaster when every county calls at once, and they solve it with labor rather than with a repeatable artifact. The gap is a narrow, opinionated, *rule-aware* document engine rather than a workflow tracker.
Proposed product
A document-first PA packet builder. Ingest: drag in the invoice PDFs, timesheets, equipment logs, contracts and purchase orders for one project. Extract line items and vendors with an LLM/OCR pipeline. Map them to PA cost categories (A–G) and to Project Worksheet line structure. Then run the part nobody else runs: a procurement compliance checker that walks each contract against the 2 CFR 200 checklist — was the method (micro-purchase / small purchase / sealed bid / proposals / noncompetitive) documented, is the exigency justification on file for a noncompetitive award, is there a cost-or-price analysis above the simplified acquisition threshold, are the required federal contract clauses in Appendix II present. Output: a submission-ready packet plus a red/amber/green exception list of every document that will draw an RFI. Explicitly NOT a portal robot in v1 — a packet builder that a human uploads. Auto-submission comes later, only after confirming the portal's terms permit it.
MVP version
Narrow it to one cost category and one failure mode: force-account labor (Category B emergency protective measures) plus the procurement checklist for contracted debris removal. Inputs: timesheet CSV/PDF, payroll rate sheet, the contract PDF. Outputs: a populated force-account labor summary, a fringe-rate calculation, and a procurement exception report citing the specific 2 CFR section for each gap. Ship as a web app with per-project pricing. One user, one county, one disaster, one packet that survives an FDEM review — that is the MVP and the only proof that matters.
30-day build
Do not write the product. Verify the two assumptions that can kill it. (1) Go to FloridaPA.org and the FEMA Grants Portal and establish, from the sites themselves, whether the FDEM portal exists as described, what its forms actually are, whether there is any API or export, and what its terms of use say about third-party tooling. (2) Establish whether Direct Administrative Costs make consultants free to the subrecipient — read FEMA's PA Program and Policy Guide on DAC and Management Costs (the 5% under Stafford Act §324). If DAC fully reimburses consultant fees, the direct-to-county channel is dead on arrival and the only viable buyer is the consultant. In parallel, use public records — the founder's actual edge — to pull FDEM subrecipient agreements and county RFP awards for 'disaster grant management services' to build a named list of which firms serve which counties and at what fee. That list is the asset.
60-day build
Build the labor + procurement-checklist MVP against real documents. Source them from public records: county commission agenda packets and FDEM's own public PA project files contain real invoices, contracts and Project Worksheets. Test the extractor and the checker against a dozen real packets and measure one number — how many exceptions it catches that a human reviewer confirms are genuine. That number is the entire sales pitch. Take it to five disaster-consulting firms from the plan_30d list, not to counties.
90-day revenue plan
Sell the checker, not the platform, to consultants as a per-packet QA pass: they upload a packet before submitting, they get the exception list, they bill the client for the review they were doing anyway at a fraction of the labour. Price per packet. Target: three consulting firms, ~$3-6k/mo combined. Simultaneously pursue one direct county pilot for the logo and the case study, priced at cost. INFERENCE, flagged clearly: 90 days to first dollar is optimistic here; 120-180 is the honest estimate because disaster-recovery buying is episodic and the founder has no existing relationship in this vertical.
Distribution path
Public-records-driven outbound — pull the FDEM subrecipient list and the county procurement awards, know exactly who is filing what before the first email, and lead with a free exception report run against one of their own already-public packets. This is 'demonstrated value, not relationship sales,' which is the founder's stated mode, and it is the only channel here that works. Secondary: the state emergency-management association conferences (FEPA in Florida) where the consultants and the county EMs are in the same room. Tertiary and slow: FEMA PA closeout/appeal chatter. There is no product-led-growth motion and no self-serve funnel for this buyer.
Pricing hypothesis
Per-packet QA review: $250-750 per Project Worksheet packet for consultants, volume-tiered. Per-seat for consultant staff: $150-300/user/mo. Direct-to-subrecipient: $1,500-3,000 per project, or a flat annual $12-25k for a county that is running many projects off one disaster. Explicitly avoid percentage-of-award pricing at first — it invites the contingency-fee framing that FEMA and state auditors scrutinise, and it makes the founder look like the consultants he is undercutting.
Technical difficulty
Moderate, front-loaded on ingestion. Document extraction from heterogeneous scanned invoices and hand-entered timesheets is the hard part and it degrades ugly — an extractor that is 90% right produces a packet that is 100% untrustworthy, so the UI has to be a human-in-the-loop verification surface, not a magic button. The compliance checker itself is the easy part: 2 CFR 200 is a finite, stable, public rule set that maps to a deterministic checklist. If the founder attempts headless submission into FloridaPA.org, difficulty jumps sharply — brittle authenticated session automation against a portal that can change without notice.
Legal / regulatory risk
Low-to-moderate and mostly about positioning. The tool must not represent itself as certifying compliance or as a substitute for legal/audit review — a deobligation is a six- or seven-figure event for a county and the founder does not want that liability. Cap liability contractually, sell it as a review aid. Handling government invoices and payroll timesheets means PII (employee names, hours, rates) and a real data-handling obligation. Per the scoring rules I am NOT flagging heavy_compliance: the founder does not need to become licensed or certified. Compliance is the moat.
Platform dependency
Deliberately structured to have none in v1. The packet builder outputs files a human uploads; it does not depend on a platform owner who can deplatform it. Per the scoring rules, no platform_policy_risk flag for a tool aimed at a government system. I am flagging large_integrations honestly: if the roadmap requires reading from or writing to FEMA Grants Portal / FloridaPA.org, that is an undocumented, likely API-less, authenticated integration and it is the single biggest engineering risk in the plan. The founder's ELDT Training Provider Registry work is direct evidence he can do exactly this — but it is one portal's worth of evidence, not a guarantee about this one.
Founder fit
Very high, and the strongest thing about this idea. Regulation compels a class to file into a government portal; a solo operator builds the submission layer; he charges per filing. That is precisely the FMCSA ELDT / Training Provider Registry shape he has already shipped and monetised. It also stacks his other edges unusually well: fire-service background means he knows what a county emergency operations centre looks like during a disaster and can talk to an EM without translation; public-records skill is the literal distribution channel here; industrial/operations background maps onto force-account equipment and labour records. The applicable lesson (0.80 confidence) — government-portal mandate opportunities are this founder's best fit — applies squarely and I am weighting it accordingly.
Breakout potential
The expansion story is genuinely strong and is FACT-supported rather than hand-waved: the same award description with the same structure appears for Puerto Rico ($35.3B), USVI ($21.99B) and New York ($17.37B). Every state runs PA pass-through; most run their own subrecipient portal on top of Grants Portal. Win Florida, and 49 near-identical markets exist with the same federal rule set underneath and only the state form layer varying. The 2 CFR 200 procurement checker — the actual defensible asset — is federal and therefore portable to every federal pass-through grant, not just disaster: transit, water infrastructure, broadband. That is a large, real second act.
Final recommendation
CONDITIONAL PURSUE — gated on one verification, not on a build. This has the exact structural shape the founder is hunting: obligated public money, a defined class that must file, a state portal, per-filing monetisation, and a fifty-state template. Founder-fit is as high as it gets. But it is an award, not a mandate change; there is no deadline; and there is a specific, checkable reason the buyer may have zero out-of-pocket cost today. Spend two weeks, not two months: confirm whether Direct Administrative Costs reimburse the incumbent consultant, and read FloridaPA.org's actual forms and terms. If DAC does NOT fully cover consultant fees, this is a strong build and the direct-to-subrecipient channel is real. If DAC DOES cover them, do not abandon it — pivot the buyer to the consulting firms and price the 2 CFR 200 procurement checker as a margin tool, accepting a smaller market with a much shorter sales cycle. Either way the durable asset is the federal procurement-compliance checker, which outlives disaster recovery and ports to every 2 CFR 200 pass-through grant in the country. Build that, and let the Florida packet builder be its first application.
Next action
Before writing any code: read FEMA's Public Assistance Program and Policy Guide sections on Direct Administrative Costs and Management Costs, and determine whether a subrecipient's grant-consultant fee is reimbursable from the award. Write the answer down in one sentence. That sentence selects the buyer, the price, and the channel for everything above.