What changed
FACT (Federal Register, OSHA, 2026-01-15): OSHA extended four HCS compliance dates by four months — 1910.1200(j)(2)(i) from Jan 19 2026 to May 19 2026; (j)(2)(ii) from Jul 20 2026 to Nov 20 2026; (j)(3)(i) from Jul 19 2027 to Nov 19 2027; (j)(3)(ii) from Jan 19 2028 to May 19 2028. FACT (same source): these are dates in the HCS final rule published May 20 2024 (89 FR 44144). FACT (Federal Register, OSHA, 2026-01-08): OSHA separately issued corrections to 'inadvertent errors' in the HCS final rule, some of which the agency itself said 'could lead to confusion during the classification process.' INFERENCE: the substantive obligation (reclassify to GHS Rev. 7, reissue labels and 16-section SDSs) is a 2024 rule; only the dates moved.
Why now
CRITICAL TIMING FACT the convergence input glosses over: today is 2026-07-10, so the first extended date — (j)(2)(i), May 19 2026 — HAS ALREADY PASSED. Chemical manufacturers and importers are already past their reclassification deadline. The live deadline is (j)(2)(ii), Nov 20 2026, roughly four months out, which INFERENCE (from the subsection structure, not asserted in the source text) governs distributors still shipping containers labeled under the prior scheme. The employer-side dates are Nov 19 2027 and May 19 2028 — 16 and 22 months out. So 'why now' is real but narrower and later than the headline suggests: the urgent buyer for the next 120 days is a distributor with old-label inventory, not a manufacturer.
Converging signals
Only one substantive signal, and the demand_evidence array is thin in a way worth naming. Of eleven FORCED BUYER items, two are the HCS rule and its corrections notice (the actual subject); the other nine are unrelated compliance-date extensions from CFPB, SEC, FMCSA, IRS, MSHA and DOL that surfaced purely on semantic similarity to the phrase 'extension of compliance dates.' They are retrieval artifacts, not converging evidence. Per the system's own lesson (conf 0.89) that lexical retrieval fabricates evidence, I am discounting them entirely: they say nothing about SDS demand. The honest signal count here is ONE rule plus ONE corrections notice. That is a mandate, not a convergence.
Customer pain
HYPOTHESIS (no complaint or hiring evidence in the input; the demand_evidence array contains zero PAIN and zero HIRING/SPEND items): a mid-size chemical blender, repackager, or distributor holds 200-5,000 SKUs, each with an SDS authored years ago under GHS Rev. 3 conventions, and has no cheap way to know which of those documents are now non-conformant or which drums in the warehouse carry a label that becomes unshippable after Nov 20 2026. The pain is inventory-scale triage, not document generation.
Who pays
The EHS/regulatory-affairs manager at a 50-500 employee chemical distributor, blender, or private-label manufacturer — someone who already carries an SDS-management line item. Secondarily, the EHS consulting firms that serve them, who would resell an audit that makes their billable hours go further. NOT the millions of downstream employers: FACT of the standard's structure is that downstream employers receive SDSs from suppliers rather than author them, so their 2027-2028 obligation is training and workplace labeling, which is a different (and far more commoditized) product.
Solved today
INFERENCE (not from the source text; from general market knowledge and therefore a claim I cannot cite): SDS authoring and management is a mature, well-capitalized paid category — VelocityEHS, 3E (Verisk), Sphera, Chemwatch, SiteHawk, KMK Regulatory. Firms either license one of these platforms with authoring modules, or retain a regulatory consultant to reclassify SKU-by-SKU. Both already sell GHS Rev. 7 conformance as a headline feature, because the underlying rule landed in May 2024, twenty-six months ago.
Why current solutions are bad
Weakly. Incumbent platforms are expensive and slow to onboard, and consultant reclassification is billed per SKU. That is a price complaint, not a capability gap. I can find no evidence in the provided sources that the existing solutions fail — and 'existing tools are pricey' is the thinnest possible wedge against six funded incumbents who ship to the same deadline.
Proposed product
Strictly an AUDIT product, never an authoring product. Upload an SDS library (PDF/ZIP or an FTP pull). For each document: parse the 16 sections, extract the hazard classification block, GHS pictograms, signal word, H/P statements, and Section 3 composition disclosure; diff against the amended HCS/Rev-7 requirement set; emit a per-SKU finding with the exact CFR citation and a severity. Deliverable is a signed PDF gap report plus a tracked remediation calendar keyed to Nov 20 2026 / Nov 19 2027 / May 19 2028. The company remediates; the tool never asserts a hazard determination.
MVP version
A parsing pipeline over a rules table encoding the amended HCS requirements, run against 200 real SDS PDFs, output as a gap report. The genuinely hard part is not the diff — it is building and defending the rules table, because OSHA published a corrections notice on 2026-01-08 admitting its own final rule contained errors that 'could lead to confusion during the classification process.' Encoding a rule the agency itself had to correct is where a solo founder's version silently goes wrong.
30-day build
Do not build. Buy 20 SDS libraries' worth of conversation: call 30 EHS managers at chemical distributors and ask what they actually did for the May 19 2026 date that just passed. If most say 'our supplier or our EHS platform handled it,' the thesis is dead and the 30 days cost nothing. Concurrently, price VelocityEHS and 3E to learn what the audit is worth as a wedge.
60-day build
Only if the calls show unmanaged libraries: hand-build a rules table for the ten most-cited HCS sections, verified against the Jan 8 2026 corrections notice, and run it against three real libraries as a paid pilot audit ($3,000-6,000 flat) with a licensed CIH or regulatory consultant co-signing the report. The co-signature is not optional — it is the entire liability structure.
90-day revenue plan
Convert two pilots to annual subscriptions covering re-audit against the Nov 20 2026 and Nov 19 2027 dates. Realistic 90-day outcome from a standing start today is $10-20k of pilot revenue, not a product business.
Distribution path
Direct outbound to EHS managers, plus co-selling through small EHS consultancies who bill the remediation the audit generates. This is the one genuinely attractive part: the audit creates consultant billable hours, so the consultant is a motivated channel rather than a competitor. Weak elsewhere — there is no marketplace, no viral surface, no self-serve funnel for a product whose buyer is one person per company.
Pricing hypothesis
$3-8 per SDS audited (one-time library scan), floor $2,500; then $2,400-6,000/yr for the compliance calendar and re-audit on each remaining deadline. The per-SDS unit is what the founder wants (per-transaction), but note it is a per-DOCUMENT fee, not a per-FILING fee — nobody is filing anything.
Technical difficulty
Moderate on the software, high on the domain. PDF parsing of 16-section SDSs is tractable. Correctly encoding hazard-classification conformance, in a standard OSHA itself just corrected, is regulatory work that the founder cannot do and cannot fully verify.
Legal / regulatory risk
Real and asymmetric. A gap report that misses a non-conformant SDS, on a chemical that later injures someone, puts the tool in the causal chain of an OSHA citation or a tort claim. Mitigable by scoping strictly to 'flag for review,' never 'certify as conformant,' and by carrying E&O — but it is a genuine reason this category is served by insured firms with credentialed staff.
Platform dependency
None. Correctly, no platform_policy_risk: there is no marketplace, no app store, no government portal that can revoke access.
Founder fit
MIDDLING, and this is where I part company with the convergence framing. The system's own heuristic (conf 0.80) is that government-PORTAL mandate opportunities fit this founder best. The input's own PORTAL field says: 'No federal portal — SDS/labels are produced and retained by the employer.' That is the ELDT pattern's load-bearing element missing. ELDT worked because a portal existed, submission was mandatory, the act of submitting was annoying, and a per-upload fee attached to a discrete event. Here there is no submission, no event, no portal, and no per-filing hook — only a document-quality obligation enforced by inspection. The founder's industrial and operational credibility helps him talk to EHS managers; his proven edge does not transfer.
Breakout potential
Low as framed. The rules table is the only asset, and it is a copyable artifact that six incumbents already maintain with staff toxicologists. There is no data network effect, and the audit is a one-time purchase with weak renewal logic once the 2028 date passes.
Final recommendation
KILL as specified; SHELVE a narrowed version. The 'SDS/label authoring and re-classification SaaS' framing should not be built: it puts a solo founder into liability-bearing hazard determination against six funded incumbents, with no portal, no per-filing event, and a headline deadline that expired seven weeks ago. The only version worth even a phone call is the strictly-scoped audit-and-punch-list product co-signed by a credentialed consultant, sold to distributors ahead of Nov 20 2026 — and that is a $10-20k consulting-adjacent business, not a sellable product with a defensible wedge. Applying the founder's own capital-and-runway lesson (conf 0.90) cuts against this idea rather than for it: he can afford a 3-6 month ramp, so he should spend those months on a mandate that has an actual portal and an actual filing, of which the Federal Register produces several per month. This is the right SHAPE (a rule compels a class to act) with the wrong MECHANISM (no submission to intercept). Recommend revisit-later only if the Nov 2026 distributor calls surface unmanaged libraries at scale.
Next action
Before writing any code: 30 discovery calls to EHS/regulatory managers at chemical distributors and blenders, asking one question — 'what did you actually do for the May 19 2026 HCS date, and who did it?' If the answer is consistently 'our EHS platform' or 'our supplier,' record the kill and move the reasoning budget to a Federal Register rule that names a portal and a submission.