What changed
On July 1, 2026 OSHA published a Federal Register notice announcing availability of $12,787,000 for FY2026 Susan Harwood Training Grant Program grants, under two separate funding opportunity announcements — Targeted Topic Training grants and Training and Educational Materials Development grants (FACT, both citations).
Why now
The FOAs are live now, which means a discrete applicant cohort is currently writing packages against a Grants.gov close date (INFERENCE: close date not stated in the provided text and must be pulled from Grants.gov before this is actionable). After the close date the pre-award half of the product has no buyer for another ~12 months. The post-award half, by contrast, turns on the day awards are announced and runs for the award period.
Converging signals
Only two signals are present, and they are the same event reported twice: the OSHA news release and the Federal Register notice announcing the same $12,787,000 (FACT). This is not three independent signals meeting at a point — it is one appropriation seen from two angles. Honest convergence here is low. What genuinely converges is structural rather than evidentiary: (a) an appropriation exists, (b) a defined class must file to reach it, and (c) two federal portals (SAM.gov/Grants.gov front end; a DOL grantee reporting system on the back end — INFERENCE, unverified) mediate every interaction.
Customer pain
HYPOTHESIS, not established by the provided sources: that small nonprofits, community and faith-based organizations, employer associations, unions and tribes lack in-house grants staff and find the SF-424 package and the post-award quarterly/SF-425/roster cadence disproportionately burdensome relative to a ~$100-160k award. No complaint threads, job postings, or Paperwork Reduction Act respondent-burden figures were provided to support this. The eligible-entity list itself is INFERENCE from the input, not quoted from the FOA.
Who pays
Pre-award: the applicant organization, out of its own funds, before any money exists. Post-award: the grantee, from indirect-cost recovery or the grant's administrative line. The post-award payer is materially better — the money is real, the obligation is non-discretionary, and the cost is allowable.
Solved today
Grant-writing consultants (fee or percentage of award), boilerplate reuse from prior-year applications, volunteer or borrowed staff time, and general-purpose grant-management SaaS (AmpliFund, Euna/eCivis, Fluxx, Foundant GrantHub, Instrumentl, Submittable) for the reporting side. Federal financial reporting (SF-425) and Grants.gov package validation are already commoditized features in that category.
Why current solutions are bad
None of the incumbents encode Harwood-specific logic: the participant/training-session roster requirements per class, the Targeted Topic vs. Materials Development distinction, or the exact evaluation criteria of the two FOAs. Generic tools give a folder and a deadline reminder; they do not tell an applicant that their workplan will score badly. That gap is real but narrow.
Proposed product
Phase 1 (post-award, the durable half): a Harwood Grantee Reporting Console — participant/session roster capture on a phone at the training site, automatic aggregation into the quarterly performance report format, and SF-425 preparation with prior-period carry-forward. Phase 2 (pre-award, the seasonal half): a package assembler that ingests the FOA PDF, produces a scored outline against the stated evaluation criteria, generates a budget narrative consistent with the SF-424A, and hard-validates the Grants.gov submission (SAM registration active, UEI present, page limits, required attachments) before submit.
MVP version
Start with the roster-to-quarterly-report path only. A web form plus mobile-friendly roster capture, a Postgres schema keyed to award/session/participant, and a report generator that emits the exact DOL-required tables and a filled SF-425. No Grants.gov submission integration in v1 — the founder's ELDT work proves he can build that later, and building it before a paying grantee exists is premature.
30-day build
Pull both FY2026 FOAs from Grants.gov and read them end to end — this replaces every inference in this brief with fact: eligible entities, close date, award size range, exact post-award reporting obligations, and which system those reports go to. Simultaneously pull the public list of FY2024/FY2025 Harwood grantees from OSHA's site; that list, not a job board, is the addressable market and it is enumerable by hand. Call fifteen of them and ask one question: who does your quarterly reporting and what does it cost you.
60-day build
If the calls surface a consistent complaint, build the roster-to-report generator against the real FY2025 reporting template and hand-run it for two prior grantees for free, producing a report they can compare to what they filed. Do not build the application assembler until the reporting side has a paying user.
90-day revenue plan
Sell an annual seat to the FY2026 award cohort the week awards are announced. Realistic target: 8-15 of roughly 80-120 awardees (INFERENCE on cohort size, derived by dividing $12.787M by an assumed $100-160k award — the award size is not in the sources) at $1,200-2,000/year, i.e. $10k-30k ARR. That is a real number and a small one.
Distribution path
The single strongest asset here is that the buyer list is public: prior-year Harwood grantee rosters and, after announcement, the FY2026 award list. Direct email and phone to a few hundred named organizations, plus presence at OSHA's grantee orientation webinars. No ad spend, no network effects. Conversely, these are small nonprofits with slow procurement-by-committee and annual budget cycles — reachable, but not fast.
Pricing hypothesis
Post-award: $1,200-2,000/grantee/year, positioned as an allowable administrative cost recoverable from the award. Pre-award: $1,500-2,500 flat per package, positioned against a consultant's $5-10k (the $5-10k consultant figure is asserted in the input, not evidenced in either source — HYPOTHESIS requiring validation on the fifteen calls).
Technical difficulty
Low. Form generation, a roster schema, PDF output. The only genuinely hard piece is Grants.gov S2S submission, and that is deferred out of v1.
Legal / regulatory risk
Low. Compliance is the product, not a barrier. The founder does not need a license or certification to prepare a grantee's report. Note: if he ever prepares the application itself for a fee he is a grant consultant, which is unregulated but does create an implicit accuracy obligation.
Platform dependency
None that can deplatform the product. Government portals do not have a policy team that bans vendors. The real dependency is on annual appropriation: if Congress does not fund Harwood in FY2027, the cohort does not renew.
Founder fit
Genuinely high on shape. This is the ELDT pattern — a federal program, a defined class that must file, a portal, a per-filing or per-seat fee — and the founder has shipped exactly that. The occupational-safety-training subject matter also sits close to his fire-service and industrial-operations background, which shortens the domain-learning curve and gives him credibility on the fifteen validation calls.
Breakout potential
The Harwood program itself will never be a large business — the arithmetic caps it around a hundred customers. The breakout case is that the reporting engine is program-agnostic: SF-425, quarterly narrative performance reports, and participant rosters are the common currency of essentially every DOL discretionary program and most federal training grants. Harwood is a beachhead with a public customer list, not a market. That is a legitimate strategy, but it means the first year is R&D funded by a token amount of revenue, and the founder should say so out loud rather than pretend the beachhead is the business.
Final recommendation
Do not build this as scoped. Do spend the 30-day plan on it, because the diagnostic cost is one week of reading and fifteen phone calls, and those calls answer a question worth far more than Harwood: does a federal training-grant recipient actually experience quarterly performance reporting and participant-roster tracking as painful enough to pay for. If the answer is yes, the correct product is a general DOL discretionary-grant reporting console and Harwood is merely the first cohort you can enumerate by name — a rare and valuable property. If the answer is no, kill it in week two having spent almost nothing. What must not happen is building the SF-424 application assembler, which serves a voluntary buyer, in a window that has probably already closed, against consultants whose fees are asserted nowhere in the evidence.
Next action
Open Grants.gov, retrieve both FY2026 Susan Harwood FOAs (Targeted Topic Training; Training and Educational Materials Development), and record four facts that this brief could only infer: the close date, the eligible-entity list, the award size range and expected number of awards, and the exact post-award reporting artifacts and the system they are submitted to. Then pull OSHA's published FY2025 grantee list and call fifteen of them.