What changed
FACT (from the provided award record): DHS obligated $2,419,828,066.60 to the STATE OF FLORIDA DIVISION OF EMERGENCY MANAGEMENT under award ASST_NON_4337DRFLP00000001_070, described as a grant to local government for repair or replacement of disaster damaged facilities (DR-4337-FL, Hurricane Irma). FACT: comparable awards exist for other grantees — $35.3B to a Governor's Authorized Representative (ASST_NON_4339DRPRP00000001_070), $21.99B to the Government of the Virgin Islands, $17.37B to New York State DHSES. INFERENCE (not stated in the input): that these are Public Assistance pass-throughs where the state is the recipient and counties, cities, school boards, special districts and PNPs are the subrecipients who actually file. Nothing in the input text states a rule change, a new deadline, or a new portal. The 'what changed' is therefore money volume, not a new mandate.
Why now
Weak. INFERENCE: DR-4337 (Irma) is a 2017 declaration; a 2026 opportunity built on it is late in that disaster's lifecycle, likely in closeout rather than intake. HYPOTHESIS: the durable 'why now' is not Irma at all — it is that Florida subrecipients structurally carry several open disasters at once (the input asserts this as inference, and it is plausible given Irma/Michael/Ian), so the multi-disaster problem persists regardless of which declaration is fresh. The input provides NO deadline ('DEADLINE: none stated in text'), which removes the single strongest urgency lever the founder's thesis relies on.
Converging signals
Three things meet: (1) very large, already-obligated federal money landing on a state agency (FACT, cited); (2) a definable class of forced filers — Florida subrecipients who must submit RPAs, Project Worksheets, quarterly reports and closeout packages to keep drawing that money (INFERENCE, the paperwork list in the input is asserted, not sourced); (3) the fact that a single subrecipient's obligations are fragmented across multiple disaster declarations with different rules, cost-share terms, and document sets. Item (3) is the only one that is actually a product insight; items (1) and (2) describe a market that has existed since the Stafford Act.
Customer pain
HYPOTHESIS, not established by the provided evidence: a county grants manager tracking Irma, Michael and Ian projects simultaneously has no single view of which project worksheet needs which document by when, and reconciliation of large projects drags for years, delaying reimbursement of money the county has already spent. The input contains ZERO pain evidence — no complaint threads, no forum posts, no job postings, no consultant marketing copy. The demand_evidence array contains only award records, including two Florida Medicaid entitlement awards (ASST_NON_2605FL5MAP_075, ASST_NON_2505FL5MAP_075) that are semantically similar but substantively irrelevant to disaster PA. That is a retrieval artifact, not corroboration.
Who pays
Three candidate buyers, in descending order of reachability for a solo founder: (a) disaster-recovery consulting firms who already manage PA claims for many subrecipients on percentage-of-award or hourly fees — they feel the multi-disaster pain most acutely, buy software without an RFP, and can be sold one firm at a time; (b) large subrecipients with dedicated grants staff (school boards, big counties, utilities, hospital districts) — real budget, but purchasing runs through county procurement; (c) FDEM itself — this is the trap: it is a single state-agency procurement, and winning it is a multi-year, RFP-gated enterprise sale. The founder's own constraints rule out (c) as a primary channel.
Solved today
INFERENCE (the input does not source this): FEMA Grants Portal and Florida's FloridaPA.org/DEMES are the systems of record. Subrecipients work inside them and keep a parallel Excel/SharePoint tracker for deadlines and document status. Larger subrecipients hire recovery consultants (Tidal Basin, Hagerty, IEM, Witt O'Brien's, CDR Maguire and similar) who bring their own internal tracking tooling. Existing grant-management vendors (eCivis, GrantVantage, AmpliFund, Tidal Basin's Recovery Cloud) already sell into this space.
Why current solutions are bad
HYPOTHESIS: the portals are per-disaster and per-project, not per-subrecipient-across-disasters; spreadsheets don't enforce deadlines; consultants are expensive and their tooling doesn't leave with the client. I cannot verify any of this from the provided sources — the entire 'why current is bad' rests on plausibility, and I am flagging that rather than dressing it up.
Proposed product
A read-and-track layer, not a submission layer. Ingest each subrecipient's project data (initially by CSV/PDF export from Grants Portal, uploaded by the customer — not by scraping), normalize it into one project register across all open declarations, and drive it with a rules engine that encodes per-declaration document requirements, cost-share terms, period-of-performance and closeout milestones. Outputs: a single deadline calendar, a per-project document-completeness score, an aged-receivable view of money obligated but not reimbursed, and an audit-ready binder generator. Deliberately NOT a Grants Portal submission bot — see technical_difficulty.
MVP version
Multi-tenant web app. (1) Upload Grants Portal / FloridaPA exports and PW PDFs; parse into projects, categories (A–G), obligated vs disbursed amounts, and dates. (2) A hand-built rules table for the three or four declarations that matter in Florida (4337 Irma, 4399 Michael, 4673 Ian, plus current-year), covering required document types per project category and per large/small project. (3) Deadline calendar with email escalation. (4) One-click closeout binder export (documents + reconciliation summary). Ship with real data from one friendly county or one consulting firm; do not build multi-state until one Florida customer renews.
30-day build
No code first. Talk to 15 people: county emergency-management directors and grants managers across Florida, plus 5 PA consulting firms. Goal is to falsify, not validate — ask what they use today, what they'd rip out, who signs, and what the purchase threshold is below which no RFP is required (in many Florida counties this is a real, findable number). Simultaneously, get read access to FloridaPA.org / Grants Portal export formats through a friendly customer. If fewer than 3 of 15 describe multi-disaster tracking as a top-three problem, stop.
60-day build
Build the ingest + rules engine for two declarations against one design partner's real project set. Instrument it: does it surface at least one deadline or missing document the customer did not know about? That single moment is the entire sales demo. Parallel: publish a free 'Florida PA closeout requirements by project category' reference on the open web — the content asset that makes the founder findable to this exact audience.
90-day revenue plan
Convert the design partner to a paid annual seat, and close 1–2 consulting firms on a per-managed-subrecipient price. Realistically, 90 days is optimistic here: government fiscal calendars, not sales skill, set the close date. Expect first meaningful revenue at 120–180 days, and expect it to come from a consulting firm rather than a county, because the firm can sign without procurement.
Distribution path
Weak, and this is the second-biggest risk after competition. Channels that work: the Florida Emergency Preparedness Association and FDEM's PA training/workshop circuit; direct outbound to named grants managers (public records make the roster trivially findable — a genuine founder strength); the free closeout-requirements reference as an inbound magnet; and reselling through recovery consultants. Channels that do not work: ads, content-for-content's-sake, and any hope of viral spread inside government.
Pricing hypothesis
Per-subrecipient annual license, roughly $4,800–$12,000/yr depending on open project count — deliberately set beneath the typical county sole-source/quote threshold so a director can buy it on a purchase order. For consulting firms, $200–$400 per managed subrecipient per month. Reject the 'per-filing fee' framing in the input: this product does not perform the filing, so there is no filing event to charge for. Charging per filing on a tool that only tracks filings is a pricing story that will not survive the first sales call.
Technical difficulty
Moderate for the tracker; severe for the version described in the input. FEMA Grants Portal has no public API, requires MFA, and is a federal system — building an automated submission or scraping layer against it is materially different from the founder's FMCSA Training Provider Registry work, where the submission path is a supported per-record upload. Treating those two as the same shape is the central technical error in the source convergence. Build the ingest on customer-initiated exports; revisit automation only if FEMA publishes an API.
Legal / regulatory risk
Low for a tracker. The data is the customer's own grant data. Two real constraints: customer-side records-retention and public-records law mean the data may itself be subject to Florida Sunshine requests, and any hosting of documentation supporting a federal award should assume audit scrutiny (single audit / OIG). Neither requires the founder to be licensed or certified.
Platform dependency
No platform owner can deplatform this — correct, per the founder's own rule. But there IS a dependency the rule doesn't cover: the product is worthless if FEMA or FDEM ships an equivalent cross-disaster view inside Grants Portal or DEMES. FEMA has been consolidating PA tooling for years. That is a roadmap risk, not a policy risk, and it is real.
Founder fit
Genuinely strong on capability: public records, systems thinking, document/deadline automation, and demonstrated government-portal work. Genuinely weak on channel: the founder explicitly avoids long trust cycles and enterprise procurement, and the natural buyer for this product is a county government. The fit is only good if the wedge is the consulting firms, who buy like businesses. Sold to counties directly, this idea sits squarely inside the founder's stated no-go zone, and I will not score around that.
Breakout potential
Moderate and honest: the rules engine, once built for Florida, ports to any state with PA subrecipients — the input's own evidence names Puerto Rico, the Virgin Islands and New York as grantees of the same program. Fifty near-identical markets is the real upside. But the same portability means an incumbent grant-management vendor can add a cross-disaster view in a quarter, and the moat is the encoded rules table, which is copyable by anyone willing to read the same public FEMA policy guide the founder read.
Final recommendation
CONDITIONAL — do not build; run the 30-day falsification first, and reframe the buyer before writing any code. The underlying observation (subrecipients carry several disasters at once and no system shows them one view) is a real and non-obvious product insight, and the founder's document-and-deadline automation skill fits it well. But the brief as generated fails on three specifics: the demand evidence is the author's own inference laundered through an award record; the monetisation shape (per-filing) is impossible because Grants Portal has no submission path he can automate; and the natural buyer is a county procurement office, which he has explicitly ruled out. The only version worth building sells to disaster-recovery consulting firms — who buy like businesses, feel the multi-disaster pain across their whole book, and already bill fees this software can undercut. If 15 discovery calls do not produce three consulting firms who describe cross-disaster tracking as a top-three problem, kill it. I am scoring this a mid-tier opportunity, not a top one, and I want to be explicit that a higher score would require pretending the evidence array says something it does not.
Next action
Pull the Florida county emergency-management and grants-manager roster from public records (67 counties, plus the top 20 school districts and special districts), and separately list every recovery-consulting firm with a Florida PA practice. Book 15 calls in two weeks. Ask one question first, before pitching anything: 'How many open disaster declarations are you carrying right now, and how do you track deadlines across them?' If the answer is 'one' or 'the portal handles it,' the idea is dead and you have spent two weeks instead of six months.