- Main Declined The board's own pricing table — its two tiers and their quotas (Starter $299 with 100 engagement letters and Professional $799 with 500), 10 and 25 seats, a 14-day free trial, and an Enterprise column
- DECLINED, not pending: every part of this table is contradicted by a decision, so there is nothing here for anybody to fill in later. Q18 was answered on 2026-09-15 and the section above publishes three real tiers, so what is declined is worth reading carefully. Two of the board's figures ($299 and $799 a month) coincide with two of the owner's prices and are attached to DIFFERENT tier names and different letter counts: the owner's tiers are Starter at $99 with 5 letters, Practice at $299 with 25, and Firm at $799 with 100. So the board's table is not a subset of the decision, it disagrees with it, and its pairing is invented sample data per design/mockups/README.md. The board's “10 seats” and “25 seats” are declined too, and for a different reason than they were: the owner reversed Q18 on seats on 2026-09-23, and the seats this product sells are Starter 5, Practice 15 and Firm unlimited — printed on the cards above from the same tier data invitations are refused by — so the board's pair disagrees with the decision rather than naming a dimension the product lacks. Storage is still not metered (Q18, 2026-09-20). The Enterprise column is declined because there is no fourth tier and there is no sales conversation to put on it (Q65). The 14-day free trial is declined because Q73 decided on 2026-09-15 that there is NO trial — the same decision as ruling out a free tier, and coherent with Q65, where an access request records a request and creates no firm. A board that draws one is drawing a product decision that was made the other way. tests/design/marketing.test.ts fails a typed seat count, the word trial, an N-day figure and an Enterprise tier anywhere this page can say one.
- Main A named firm in the strip below the hero, and the heading that would go over one
- The strip is built since 2026-09-22 and the list of firms behind it is empty. The board draws five client-logo placeholders under a line asserting that those firms run on this product, and no firm has ever been onboarded: app.firm is provisioned by Selantia staff (Q65). So the strip draws empty wireframe blocks and calls itself a layout placeholder in its own words, and the heading that names firms lives on a branch src/marketing/proof.ts cannot reach while no firm has consented (Q15, D8). This entry was marked as decided-against until 2026-09-22, when the owner asked for the section: what was decided against is the claim, and the claim is still not made — what arrived is a strip that cannot make it. An entry needs a real firm and that firm's consent to be named, which since 2026-09-22 is recorded in the product by Selantia staff, audited and withdrawable (Q97), and covers the name only. No firm has been onboarded, so none has consented, and the strip is empty.
- Main A real firm's estimate queue in the hero frame, and any column in it that states a tax outcome
- The frame is built since 2026-09-22 and what it shows is sample data, labelled as that on the page in the board's own words. What is absent is a real one, and it is absent permanently rather than pending: a screenshot of this product shows a real firm's clients, which is the thing the whole repository is built to keep apart (Q20), and Q47 lets a FIRM publish an example of its own work because that is a claim with an author. So the rows carry no client, no city and no firm: the three labels are not names at all, each basis is an exact multiple of a million dollars so that no reader takes one for data, and the statuses are Q17's own five imported from the product rather than written here. Also absent, and this half is a decision: a benefit, savings or deduction column. A basis is a fact about a property that a firm types in; what a study will find is a tax outcome, the bonus mechanic is uncited and eligibility is the firm's judgement (Q41, D8, Q4). This entry was marked as decided-against until 2026-09-22, when the owner asked for the frame, and the argument the previous pass deleted it on was half wrong: a product illustration with clearly-fictional data asserts nothing about who uses the product, which is a different claim from a wall of logos.
- Main “Talk to us” and “Contact sales”
- Q65 answered the primary call to action and nothing else. These are a sales conversation rather than a recorded request — a different destination and a different promise — and pointing them at the same form would be deciding they are the same thing. The footer's contact and support addresses were supplied by the owner on 2026-10-04 and are linked, and the LinkedIn page and SwiftSeg's website were supplied on 2026-10-05 and are linked too, so they left this entry (Q65). components/tenant/landing.tsx made the same call about “Talk to an engineer”.
- Main Anything the request-access form promises about what happens next, and any firm-size field
- Q65 asks what a form may promise and the answer is nothing nobody has committed to: “we reply within a day” and “live in days” are service levels nothing measures. ⚠️ The acknowledgement was a bracketed slot until 2026-09-22 and is written now (Q15, answered again on 2026-09-21), and what it says is what happened — the request is recorded, no account was opened, there is nothing further to do. No time, no next step and no promise, and the test suite fails a timing claim anywhere on either page. The board's flow also implies a firm size, which is only useful as a bucket and a bucket is a taxonomy this product does not have — the argument app.property.building_type already makes about a dropdown of property types. A free-text note carries it instead.
- Main Final, lawyer-reviewed Privacy, Terms and Cookies pages
- Q95: the three pages exist since 2026-10-04 as drafts the owner asked the agent to write from what the code does, and each says on the page that it has not been reviewed by a lawyer. Whether they are sufficient is a lawyer's question. The owner supplied the legal entity, the governing law and the effective date on 2026-10-05, and decided there is no postal address; the courts on 2026-10-06 (Delaware). The liability text is still a marker. Nothing on them claims compliance with anything.
- BrandSibling SwiftSeg's lime on the parent badge — --color-swiftseg, which the board reserves for the parent brand
- D24 and Q43: Main.dc.html's ramp wins over BrandSibling's, and Main draws the badge in ink on paper. --color-swiftseg also has no row in src/design/accessibility.ts — 8a measured the roles the foundations define and this is not one — so lime text or a lime fill under a label would be an unmeasured pair. Adding one needs a measurement, not a decision to use it.
- Mobile Any small-viewport behaviour at all
- D7 and Q51: Mobile.dc.html draws the estimates list and the estimate detail, and nothing else. No breakpoint is extrapolated here any more than elsewhere. ⚠️ This is the page in this product most likely to be opened on a phone, which is recorded in 22-marketing.md as a fact rather than as a licence to guess a threshold.