Long-list matrix · 45 products · 5 checks · 5 groups

Picking the 3–5 we actually work on.

Every viable product in the catalogue scored against five plain-English checks — same direction throughout, so green always means good for shipping. Pre-read for the Rob meeting on Wednesday 3 June.

v0.2 · redesigned for clarity 45 products · 18 parked excluded

How to read this

Each product is scored on five checks. Every column uses the same traffic-light: green = passes, amber = partial, red = fails. You don't need to remember which direction is "good" for each column — green is always good.

Passes — good for shipping on this dimension Partial — works with mitigation Fails — blocked on this dimension

The ★ star marks the 5 shortlist products. Rows are grouped by verdict — what kind of product it is — not by score. Inside each group, sorted by commercial attractiveness.

The five checks:

Stakes — bearable? If AI gets it wrong, is the worst-case bounded? Green = yes (£100s, reversible). Red = no (irreversible / decades-delayed / kills the brand).
Accuracy — can AI do it? Today's frontier models, with our pipeline. Green = yes (>95% on a clean test set). Red = no (silent structural errors).
Verifiable — can we tell if it worked? Green = auto-confirmed in days/weeks (Sierra-shape). Red = unverifiable until something happens years later (wills, LPAs).
Reg clear — can we ship today? Green = LSA-clear & FCA-clear. Amber = sector watchpoint. Red = conduct-of-litigation grey or reserved instrument.
High demand — is the market big? Green = top tier in the research register (score ≥ 25/35). Amber = middling (20–24). Red = small (< 20).

§ 1 The matrix

45 products, grouped into 5 verdict types. The 5 shortlist picks (★) sit across groups A, C and B — they're a composed narrative, not just the rows with the most green.

A · Outcome-pricing pilots12 products · Sierra-shape

Low stakes, AI does it well, the outcome is auto-verified within days or weeks. These are the products where we could credibly pilot pay-only-if-it-worked pricing — the model Ed is pulled towards.

#	Product	Score /35	SL
01	Parking-ticket challenge28-day operator response, then POPLA/IAS	28	★
02	GDPR Subject Access Request30-day statutory deadline, auto-verified	25	★
06	Flight delay claim (EU261)Airline pays or refuses within weeks	25	★
07	Faulty goods / S75 letterRetailer or card issuer responds	25	☆
04	Tenancy deposit checkLandlord returns or escalation to scheme	23	☆
08	Energy bill complaintSupplier responds, then Ombudsman	23	☆
21	Council Tax band checkVOA rebanding decision with rebate	22	☆
29	Consumer S75 / chargeback packComposite letters across the claim path	22	☆
59	SAR bundle (top 20)20 SARs issued in parallel	22	☆
03	GDPR Article 17 erasureSame shape as SAR, lower volume	21	☆
09	Telco cancellationConfirmation email or Ofcom escalation	19	☆
62	Subscription cancellationConfirmation email or refund	18	☆

Group A reads as: 12 products in the Sierra-shape zone. Three picked for the shortlist by motion diversity (defence / information / claim) plus high demand. The other 9 are bench depth: if the model works on pilots 1–3, this is what extends next.

B · Brand-flywheel & funnel products14 products · Mode 1 ship-as-tool

Documents the user takes away and uses themselves — casual contracts, diagnostics. Stakes mostly medium (artefact sits in drawer until tested), accuracy good, verifiability low (you only find out the contract was weak when a dispute happens). Right shape under the Rinder brand-trust umbrella as funnel / membership-stickiness products — not standalone scale businesses.

#	Product	Score	SL
40	Builder mock-up contractPlain-English homeowner–builder contract	30	★
41	Builder variation-order toolCompanion to #40 · mid-job amendments	29	★
14	Personal loan agreementFriend–family · outside FCA regime	28	☆
17	Will Checker (G2)Free 5-min diagnostic → Wills G3 funnel	28	☆
19	Estate Planning Readiness8-dimension diagnostic → Bundle	27	☆
13	IOU / Promissory NoteSingle-signature debt acknowledgement	25	☆
42	Freelance / Contractor AgreementB2B PLG · self-employed audience	25	☆
55	Neighbour dispute packComposite letters · Rob endorsement fit	25	☆
05	Section 21 validity checkDiagnostic + tenant defence letter	25	☆
61	Personal Legal MOT6-domain diagnostic · widest funnel	25	☆
18	LPA Readiness (G2)LPA diagnostic → LPA G3 funnel	24	☆
15	Private car sale agreementBill-of-sale style e-signed	19	☆
16	Lodger agreementExcluded-occupier licence	19	☆
43	NDAMutual confidentiality agreement	16	☆

Group B reads as: the brand-flywheel cluster. Two shortlist picks — Builder mock-up + Variation-order — because they're top-of-portfolio by commercial score AND they're the cluster Rob raised on 19 May. The diagnostic feeders (17, 19, 61, 18) are best supporting cast; the rest are fill-out-the-membership-bundle.

C · Estate cluster — needs solicitor review7 products · Mode 2 HIL

High stakes (failure delayed by decades, affects third parties), reg-clear, high demand — but verifiability is structurally low. The solicitor-review layer is what makes them viable. This is the cornerstone business in the brand-trust thesis.

#	Product	Score	SL
25	Wills (G3)£79 solicitor-reviewed · cornerstone	29	★
27	Estate Planning BundleWill + 4 LPAs + Roadmap · £249–£399	27	☆
53	SEND tribunal packSch.2 para.4(2) tribunal carve-out	27	☆
26	Lasting Power of Attorney (G3)£99 single · OPG checkpoint built in	26	☆
47	CodicilWill amendment · mini-will failure mode	21	☆
50	Letter of WishesNon-binding companion to Wills	21	☆
51	Funeral preferencesNon-binding companion to Wills	20	☆

Group C reads as: the cornerstone cluster. Reds on Stakes and Verifiable are the structural ones — they don't disqualify the products, they explain why solicitor review is part of the offer, not a cost line. One shortlist pick: Wills G3. The Bundle (27), LPA (26) and SEND pack (53) are the natural cross-sells once Wills is established.

D · Reg-blocked today6 products · pass framework, fail reg

Court-form work in litigation-grey territory (the CRR cluster) plus a couple of FCA-watchpoint products. All pass the Verifiable Frontier framework. None can ship today without counsel sign-off — or, for the CRR cluster, without Pablo's pop-up form-window pattern resolving the conduct-of-litigation question.

#	Product	Score	SL
39	County court triage (reframed)Diagnostic for defendants · CRR	29	☆
35	County Court Defence (CCD)Defendant-side defence drafting · CRR	28	☆
36	CCJ set-asideApplication to overturn default judgment · CRR	27	☆
37	Directions QuestionnaireProcedural form · CRR	24	☆
23	Employment grievance (G2)FCA watchpoint · ACAS reframe pending	25	☆
54	DWP / PIP appealFCA watchpoint · specified benefit	24	☆

Group D reads as: the biggest unlock sitting outside today's shortlist. The CRR cluster covers 1.7m county-court claims a year and 90% currently go undefended — one of the largest access-to-justice surfaces in the UK. Pablo's form-window investigation is the path to moving these into shortlist territory. Worth flagging to Rob as "this is the next big move if Pablo's technical pattern lands."

E · Fails framework as designed1 product

The Estate Kit's stated design — high-stakes template product distributed at consumer scale via partner channels, no solicitor review — combines the worst of the estate cluster's stakes with none of the safeguards. Partner distribution amplifies the brand-blast-radius rather than buffers it.

#	Product	Stakes	Accuracy	Verifiable	Reg clear	High demand	Score	SL
60	Estate Kit (referral)Partner-distributed · explicitly no solicitor review						27	☆

Group E reads as: the one product that needs redesign before partner-scale shipping. Three options: (a) add a lighter solicitor-review tier at partner cost; (b) reposition as explicitly preliminary with hard upgrade-recommendation; (c) confine distribution to lower-risk segments only. Charity-channel distribution is the canon's biggest single mechanism (Farewill precedent), so resolving the design is more valuable than parking the product.

§ 2 The shortlist — 5 products to put to Rob

Composed as a coherent narrative, not the top 5 by score. Three outcome-pricing pilots in three different verticals + the Mode 2 cornerstone + the casual-contracts flywheel. Three motions, one business.

#01Parking-ticket challenge

Outcome pilot · vertical 1

Low stakes AI does it Auto-verified Reg clear High demand

The cleanest product in the portfolio. £25–£170 stakes, 28-day auto-verified outcome via the operator or POPLA/IAS, 12–14m PCNs issued annually in the UK. Already shipped in private beta as parking-defence.

Why this leads. The product that lets us pilot outcome-pricing (pay only if cancelled) before betting the company on it in higher-stakes domains. Ed in the huddle: "that's the verifiable success." Brand-builder, traffic-magnet, and the operational sandbox for Sierra-shape economics.

#02GDPR Subject Access Request

Outcome pilot · vertical 2

Low stakes AI does it Auto-verified Reg clear High demand

Free template letter forcing a data controller to disclose held data within 30 days under UK GDPR. Statutory backbone, auto-verifiable response, no FCA/LSA surface. Pairs naturally with debt-defence flows and consumer disputes.

Why this earns the slot. A second outcome-pricing candidate with a different motion shape from parking — information request, not defence. Demonstrates the framework works across multiple problem types. Also a natural funnel into the SAR bundle (#59) and the debt cluster.

#06Flight delay claim (EU261)

Outcome pilot · vertical 3

Low stakes AI does it Auto-verified Reg clear High demand

Letter to airline asserting EU261/UK261 compensation entitlement (£220–£600 per passenger). Highly templated legal terrain — statutory grounds — auto-verified by the airline paying or refusing. High demand intensity.

Why this earns the slot. Third outcome-pricing pilot at higher per-transaction value than parking. Existing market (AirHelp, Bott & Co) proves willingness to pay 25–30% of recovery for the convenience — we sit underneath their pricing as self-serve. 1+2+3 give us three independent outcome pilots in three verticals before we touch anything that needs solicitor review.

#25Wills (G3) — with #17 Will Checker as feeder

Mode 2 cornerstone

Stakes (HIL bridges) AI mostly does it Verified at probate Reg clear High demand

£79 solicitor-reviewed Will, drafted by multi-agent AI pipeline from a plain-English interview. Will Checker is the free 5-minute diagnostic that funnels into it. 33m UK adults without a will. Existing pure-play competitors prove the unit economics work when bundled with adjacent products.

Why this earns the slot. The outcome-pricing pilots are revenue-positive sandboxes — this is the business. The Mode 2 cornerstone the brand-trust thesis is built on. Solicitor review is the layer that bridges the verifiability gap; it's part of the offer, not a cost line. Pair shipped together because the funnel mechanic is the product.

#40+41Builder mock-up + Variation-order

Brand flywheel

Medium stakes AI does it Sit-in-drawer Reg clear High demand

Top-scoring sorted product (Builder mock-up, #40) plus its companion (Variation-order, #41). Free first contract, £7.99/mo or £49/yr unlimited. The casual-contracts cluster anchor — the product Rob raised on 19 May as the opportunity-signal.

Why this earns the slot. The framework calls this consumer Mode 1 — structurally marginal as a standalone scale business. But as the brand-flywheel under the Rinder umbrella, it's the funnel-and-stickiness engine. Highest commercial score in the portfolio. Sits cleanly in the free / membership economics. The "every household has one of these moments" front-of-funnel product.

What the five give us together. Three outcome-pricing pilots in three verticals (#01 #02 #06) — learning Sierra-shape economics on low-stakes auto-verifiable domains, with quick revenue. One Mode 2 cornerstone (#25 + #17) — the brand-trust play that pays for the platform. One Mode 1 brand-flywheel (#40 + #41) — the casual-contracts engine that builds awareness and recurring touch. Three motions, one coherent business.

§ 3 What the matrix doesn't tell you

Four points sitting alongside the matrix that Ed and Rob should weigh in the meeting.

The CRR cluster (Group D) is the biggest unlock outside the shortlist. All four court-form products pass the framework but are reg-blocked today. Pablo's pop-up form-window investigation is the path. If that pattern works and a regulatory lawyer signs off, the entire defence cluster (1.7m county-court claims/year, 90% currently undefended) moves into shortlist territory.
Estate Kit (#60) is portfolio-strategic, not product-strategic. The product isn't broken — the partner-distribution-without-HIL design is. Resolving it (rather than parking) matters because charity-channel distribution is the biggest single mechanism in the canon — Farewill built on 400 charity partners + £1.1bn pledged charitable income.
The accuracy axis is a snapshot. Per the canonical framework's frontier-test cadence, today's scores need re-running after each meaningful frontier model release. Wills only crossed the Mode-2 threshold recently. Codicil (#47), Lodger (#16), and the Roadmap component of the Bundle (#27) sit close to the boundary — one or two may cross next cycle.
What's not here. The 18 hygiene-fail products from the source register are excluded. The B2B/SME cluster is parked pending sub-brand decision (Group C in the source). Reserved-instrument deeds are out of scope under the LSA perimeter.

Draft status. v0.2 for Rachel review. Three Tests scoring for the 33 new products is my best read against the briefs and the canonical framework — not yet panel-validated. The shortlist convergence is robust to most plausible re-scoring (none of the five flip with one-step changes).