The substrateY sylwedd
Each layer is fetched from its statutory publisher, clipped to the local bounding box, and held in a local spatial cache. Welsh-language content from bilingual publishers is preserved as-is.
Map is illustrative — substrate volumes for each layer are shown below in measured counts.
Layers cached to dateHaenau a gadwyd hyd yma
Honest status per layer. Some are fully cached; others queued; some report zero because the underlying signal is genuinely absent (the substrate's measurement, not a failure).
HMLR Price Paid
Prisiau Tir y Goron
6,185
recorded sales 1995–2026
Crown · OGL · Brief 1 landed
English-only publisher
EPC Register
Cofrestr EPC
4,074
certificates with detail · 83.2% postcode coverage
gov.uk · OGL · Brief 2 landed
Welsh lodgements preserved as-is (~10 certs)
INSPIRE Polygons
Polygonau Cadastraidd
15,758
cadastral parcels · 86.46% UPRN coverage
HMLR INSPIRE · OGL · Brief 3 landed
English-only publisher
Flood Risk
Risg Llifogydd
159,582
flood-risk polygons across 7 datasets
NRW / Cyfoeth Naturiol Cymru · OGL · Brief 4 landed
Welsh + English fields preserved
Coal Mining
Mwyngloddio Glo
0
features in LL bbox — Gwynedd is outside the coal reporting area
MRA · OGL · Brief 5 landed
Substrate signal: confident absence
Environmental
Amgylcheddol
queued
UKHSA radon · BGS ground hazards
Brief 6 · ready for Code · Friday 22 May
Awaiting ingestion
Buildings
Adeiladau
queued
OS building footprints · per-structure granularity
Brief 7 · ready for Code · OGL discipline
Awaiting ingestion
Planning History
Hanes Cynllunio
queued
Eryri NPA + Cyngor Gwynedd portals
Brief 8 · ready for Code · statutory-public
Expected bilingual at source
Cadw
Cadw — Hen Eirf
queued
listed buildings · scheduled monuments · parks & gardens
Brief 9 · ready for Code · OGL
Strongest Welsh content in substrate
Council Tax Bands
Bandiau Treth Cyngor
queued
VOA per-UPRN bands · Welsh A–I taxonomy
Brief 11 · drafted · OGL
Band letters universal; descriptions English-only
UPRN positions
Safleoedd UPRN
9,589
cached locations across LL40 / LL35 / LL36
Lleolydd Phase 1 · operationally live
Address-based identifier
Linkages
Cysylltiadau
queued
candidate cross-links · curator review queue
Brief 10 · runs after Briefs 7+8+9
Surfaces to Lleolydd iPad PWA
Framework
Fframwaith
v0.1.0
awen-weave · live on PyPI · AGPL+commercial
Awen Weave Limited 17229089
Bilingual schema vocabulary
Cross-reference measurementMesur croesgyfeirio
Phase A.5 cross-references HMLR sales against EPC certificates and other substrate signals to auto-pin transactions to specific buildings where possible. Honest measurements; predictions and reality have diverged repeatedly through the build-out.
Confident-match percentage within the HMLR + EPC matchable set. Each version adds a substrate signal: v1 used four documentary signals; v2 added cadastral polygon shape; v3 will substitute building-footprint polygon shape; v4 will incorporate curator-confirmed linkages. The headline is not the raw percentage but the cumulative shift in where the substrate has confidence and where the curator's eyes are needed.
Substantive finding · v2
Welsh-rural cadastre doesn't fit English-urban property taxonomy
Of 9,589 cached UPRNs measured against INSPIRE cadastral polygons, polygon-shape inference classified 49.55% as flat and 31.49% as complex / other — only ~19% fit the standard detached / semi / terrace / flat taxonomy.
This is the substrate working correctly: Welsh-rural parcels routinely encompass multiple buildings (courtyard, longhouse, chapel-conversion, vernacular cottage). Brief 7 ingests OS building footprints to give the per-structure resolution the heuristic was designed for; v3 re-measures.
Sail — and what's in the packSail — a beth sydd yn y pecyn
The name · Yr enw
Sail works in both languages. In Welsh, sail means foundation, basis, the ground beneath — "ar sail tystiolaeth", on the basis of evidence. In English, sail is the means of voyaging. Same word; both meanings active. The Welsh meaning names what the substrate is. The English meaning names what the product does.
Sail through the purchase. Smooth voyage in your new home.Sail i'ch prynu. Mordaith hwylus yn eich cartref newydd.
Sail assembles two artefacts from the Town Dataset substrate. A transaction pack at sale — pre-filling the solicitor's TA6 disclosure with cited sources, attaching the certificates that compliance requires, and surfacing what the owner must still declare. And a living record after sale — accumulating new certificates, tracking renewal cycles, and updating as the substrate evolves around the property.
Substrate
Sylwedd
Layers cached from statutory publishers. Bilingual where the source publishes bilingually. Honest counts; honest absences.
Candidates
Ymgeiswyr
Substrate produces calibrated-confidence candidate linkages: which UPRN belongs to which building, which Cadw record protects which structure.
Curator
Curadur
iPad PWA. Curator walks the street, reads bilingual descriptions, confirms, rejects or corrects candidates. Seconds-per-decision, not hours.
Sail pack
Pecyn Sail
Transaction pack at sale, then a living record across ownership. Cited sources, bilingual content, curator provenance.
Sample pack · illustrative
10 Bridge Street, Dolgellau
10 Stryd y Bont, Dolgellau · LL40 1AB
Identity
UPRN ending …1234 · INSPIRE parcel containing 2 building footprints · curator confirmed: front terrace house, principal structure
Heritage · English (Cadw)
Grade II listed. Mid-18th-century stone-faced terrace house with rendered front and slate roof. Within Dolgellau Conservation Area.
Treftadaeth · Cymraeg (Cadw)
Wedi'i restru — Gradd II. Tŷ teras canol y 18fed ganrif gyda blaenwedd o garreg wedi'i rendro a tho llechi. O fewn Ardal Gadwraeth Dolgellau.
TA6 · pre-filled from substrate (with citations)
TA6 · owner declares (with substrate prompts)
Certificates attached (owner-supplied)
EICR · 2024-03 Gas Safety · 2025-08 FENSA · front sash · 2018 Building Control · rear extension · 2014 Listed Building Consent · 2014 Roof guarantee · 2019 · 25-yr Cadw conservation grant · 2017Substrate signals at a glance
Grade II listed Conservation Area Flood risk · low Coal mining · no concern EPC band D · 2021 Council Tax · Band D 3 historic transactionsLiving record · renewal reminders & substrate updates
Curator notes (bilingual; signed by curator)
"Tŷ blaen yn unig — heb gynnwys yr estyniad cefn modern." · "Front house only — modern rear extension excluded from the listing."
TA6 pre-fill — the load-bearing claim
Every UK home sale requires a TA6 Property Information Form. Owners currently fill it manually, often inaccurately, with thin provenance. Sail pre-fills the items the substrate can answer — with cited sources from statutory publishers — and surfaces the rest as owner-declared with substrate prompts to guide accuracy.
Living record after sale
The pack doesn't go in a drawer. It accumulates new certificates as the owner adds them, surfaces renewal reminders, and updates as the substrate evolves around the property (new planning applications nearby, NRW reassessments, Cadw changes). The pack you hand to your next buyer is richer than the one you received.
What stays commercial
Local Authority CON29 (drainage / road adoption / formal LA enquiry) — substrate does not displace these. Chancel repair indemnity is an insurance product, not data. PI-insured commercial wrap stays above the substrate.
Cost framing
Substrate-displaceable layers across a typical residential pack: ~£100-200. Plus the value of TA6 disclosure that's faster, harder to make mistakes on, and stays current. Plus the post-sale living-record dimension that the one-shot transactional model has never offered.
Together with Olive — from substrate to buildingGydag Olive — o sylwedd i adeiladu
A proposed three-layer partnership stack. Awen Weave holds the substrate; Sail packages it for the homeowner across the ownership lifecycle; Olive turns it into construction action — refurbishment specifications, complete contracts, automated tender, and contract insurance. Three layers; one provenance trail; cited sources throughout.
Action layer
Olive
Adeiladu — Construction services
Takes the agent's house plans and the substrate's documented property profile to produce structured construction outputs:
- Quick refurbishment specification generated from plans + substrate constraints (listed status, planning history, conservation area, EPC profile)
- Alteration scope informed by the property's actual building footprint + curator-confirmed structural identity
- Complete contracts tied to the agreed scope, with provenance back to the substrate sources
- Basic management package — milestones, sign-offs, document store
- Automated tender to local trades, with substrate-aware scoping
- Contract insurance for construction works
Journey layer
Sail
Pecyn a chofnod byw — Pack and living record
Packages the substrate for the homeowner — both at purchase and across the years of ownership:
- TA6 pre-fill at purchase with cited statutory sources
- Certificates & warranties attached as the owner adds them (EICR, Gas Safety, FENSA, BC sign-offs, Cadw consents)
- Renewal reminders surfaced on the right cadence (annual gas, 5-yr EICR, 10-yr EPC)
- Substrate-evolution notifications (planning applications nearby, NRW reassessments, Cadw designation changes)
- Provenance trail handed to the next buyer at re-sale
Substrate layer
Awen Weave
Sylwedd a sail — Substrate and foundation
The open, governed data substrate. Bilingual at the data layer; honest about what's there and what isn't.
- OGL data layers cached from Crown, Welsh Government and local-authority publishers
- Curator-confirmed linkages via Lleolydd iPad PWA — substrate produces candidates; curator confirms in the street
- Welsh-language content preserved at source; bilingual records carry both languages structurally
- Forward-additive claim model — record persists across ownership transitions; new owners add, no-one deletes
- Awen Weave framework available under AGPLv3 + commercial dual-license; instance data under OGL
What the partnership earns · Gwerth y bartneriaeth
For the homeowner: a continuous, trusted path from "I'm buying this house" through years of ownership to "I have an insured contract for the refurbishment" — built on the same substrate, with one provenance trail, in one or both languages.
For the substrate: validation that the data being captured genuinely serves downstream construction action, not just compliance. Every TA6 disclosure that helps Olive scope a refurbishment is substrate doing real work.
For Olive: a property-aware starting point for any construction engagement. The substrate already knows the listed-building status, planning history, conservation-area constraints, EPC profile and building footprint. Olive starts from "here's everything documented about this property, with cited sources" rather than "here's a plan, we'll figure out the rest."
For Welsh-rooted credibility: a partnership stack that operates bilingually end-to-end. The TA6 reads in Welsh for the Welsh-speaking buyer; the substrate carries the Cadw description in Welsh; the Olive refurbishment specification cites its sources in Welsh where the source published in Welsh. Substrate-bilingualism propagates upward through the stack rather than getting flattened at the surface.
Bilingual at the data layerDwyieithog yn yr haen ddata
The substrate does not normalise Welsh-language content to English. When a publisher carries Welsh fields, the substrate preserves them; when a buyer reads their pack in Welsh, they read the original-publisher Welsh text, not an after-the-fact translation.
The disciplineY disgyblaeth
Three things are true about how Awen Weave handles Welsh-language data:
Don't translate. When NRW publishes a flood-risk descriptor in Welsh, that Welsh descriptor sits in the substrate and surfaces in pack content unmodified. The Welsh-language buyer sees the Welsh authority's original Welsh phrasing.
Don't pick a primary language. Bilingual records carry both languages side-by-side; downstream code decides what to render based on reader preference.
Don't fabricate where absent. Where a publisher carries English only, the substrate carries English only. No back-translation; no inferred Welsh.
Coverage by sourceSylw fesul ffynhonnell
- Cadw · listed buildings + monuments100%
- NRW · flood-risk + environmental100%
- Eryri NPA · planning history~95% (expected)
- Cyngor Gwynedd · planning historyvariable (Idox install)
- EPC · gov.uk register~10 Welsh certs preserved
- HMLR INSPIRE · cadastral parcelsEnglish only at this layer
- MRA · coal miningEnglish only
- OS · building footprintsEnglish only
- VOA · Council Tax bandsband letters universal