What changed
FACT: In mid-2026 the FDA published rules classifying radiological ML-based quantitative imaging software with a Predetermined Change Control Plan into class II with special controls (Fed. Reg. 2026-12166, published 2026-06-17), alongside other special-controls classifications (e.g., 2026-12443). HYPOTHESIS: this opens a class II special-controls path that small teams can pursue without de novo/PMA-scale budgets — the source confirms only that the classifications now exist, not that small teams can self-file cheaply.
Why now
HYPOTHESIS: the classifications are only weeks old, so early entrants are drafting submissions now with little productized precedent, and the window closes once incumbent regulatory consultancies package their own PCCP templates. This 'why now' is inferred; no evidence of actual small-team submission activity is provided.
Converging signals
FACT: two+ 2026 FDA special-controls classification rules in the input. INFERENCE: everything connecting these to a paid scaffolding product (consultancy pricing of $300–500/hr, seed-team preference for fixed-price kits) is asserted in the convergence description without a cited source and must be treated as unproven.
Customer pain
HYPOTHESIS (unproven): seed-stage device-software founders find regulatory drafting a bottleneck and would pay for scaffolding. The input's demand_evidence contains ZERO complaints, job posts, or founders describing this pain — only the classification rules themselves. Demand is asserted, not evidenced.
Who pays
Hypothesized: pre-seed/seed radiology-AI and remote-monitoring device-software startups (their founders or fractional regulatory leads). This is a small, hard-to-identify, hard-to-reach buyer pool with episodic (once-per-submission) need.
Solved today
FACT (general knowledge, not from source): regulatory consultancies and fractional RA/QA consultants; FDA's own guidance documents and the eSTAR electronic submission template, which already structures 510(k)/De Novo content including PCCP sections.
Why current solutions are bad
Claimed: hourly consultants are expensive ($300–500/hr) and enterprise-shaped. INFERENCE — but the same consultants also carry the liability, judgment, and FDA relationship that a static template pack cannot, which is precisely what a first-time filer is buying.
Proposed product
A per-submission 'scaffolding' pack: annotated PCCP templates, special-controls conformance checklists, and V&V protocol skeletons compiled from public FDA text and cleared-predicate summaries, sold at $2k–$10k.
MVP version
Compile the codified special-controls text for the new device types + FDA PCCP guidance into one annotated, fill-in template pack; build a simple landing page; attempt cold outreach to 20 seed device-AI startups (the convergence's own test).
30-day build
Run the falsification test FIRST before any build: cold-email/LinkedIn 20 seed radiology-AI and remote-monitoring startups offering the fixed-price pack; book discovery calls; ask directly whether regulatory drafting is their bottleneck and whether counsel already covers it cheaply.
60-day build
Only if ≥1 LOI and founders confirm the pain: build the template pack with a regulatory contractor to ensure accuracy (accuracy liability is the whole product).
90-day revenue plan
Sell 1–3 packs at $3k–$8k. Realistically first revenue is unlikely inside 90 days given device-team submission timelines run 6–18 months and buyers are scarce.
Distribution path
Cold outreach + RAPS forums + LinkedIn to a narrow, hard-to-enumerate founder list. No efficient channel; each buyer needs individual education and trust.
Pricing hypothesis
$2k–$10k per submission pack (hypothesized willingness-to-pay; unvalidated).
Technical difficulty
Low technically (documents), but HIGH domain-expertise difficulty — the value is regulatory accuracy the founder does not possess and would have to buy.
Legal / regulatory risk
HIGH. Selling regulatory submission content to medical-device makers invites liability if a template is wrong and a submission fails; borders on unlicensed regulatory advice; medical-device space is heavily regulated.
Platform dependency
Low direct platform dependency, but total dependence on evolving FDA guidance and eSTAR, which can render a static pack obsolete or trivial.
Founder fit
POOR despite surface resemblance to his FMCSA portal win. His proven edge is per-filing automation against a government portal for a large, clearly-forced, easy-to-reach class of filers (truck-driver trainers). This is the opposite: a tiny, hard-to-find buyer pool, no forced deadline, no portal automation (it's static documents), and it sits squarely in 'heavily regulated medical products' — an explicit avoid. He lacks the regulatory-affairs credibility buyers require.
Breakout potential
Low-to-moderate; even if it works it's a boutique document business, not a scalable per-transaction portal tool.
Final recommendation
KILL (or at most, spend one week running only the free falsification test). This fails on demand (no real evidence, misclassified forced-buyer signal), founder fit (explicit medical-regulated avoid, expertise he lacks, liability), and defensibility. It resembles his FMCSA win only superficially — there is no portal to automate and no compelled class of filers with a deadline.
Next action
Do NOT build. If curious, spend under a week on the zero-cost test: message 20 seed device-AI founders and ask whether regulatory drafting is their bottleneck and whether counsel already covers it — kill immediately unless ≥1 unambiguous LOI appears.