Free · 2 credits for the AI verdict
Draw a line. Get an answer.
An appraisal is the whole question answered at once: what is on your site, how likely it is to go through, what your council is actually like, which documents you will need, and what the whole thing costs.
Rendered live
This panel is the real thing.
Not a screenshot. The verdict, the likelihood and every factor beside it are computed on this page by the same functions the product runs, and the authority's figures are read from the same table.
The constraint set is a worked example, spelled out below. What is computed from it is live, which is why the numbers here move when Westminster’s do.
The likelihood starts at the national baseline of 86%and is moved by named factors, each with a signed number. A conservation area costs ten points. Flood Zone 3 costs fifteen. A listed building nearby costs eight. Your council’s own record moves it again, up or down. You can see every step, which means you can disagree with any of them.
What the appraisal never does is invent a figure to fill a gap. A field we do not hold prints as not held. It does not print as zero, and it does not quietly borrow a national average and present it as your council’s.
Reading live authority figures…
A householder rear extension on a Westminster terrace in a conservation area, inside the tidal flood zone. Constraints are a worked example; verdict, likelihood, factors and authority figures are computed live.
What runs
Seven things happen when you finish the line.
Each one is separately timed out, so a slow government endpoint delays one row rather than the whole answer.
01
Everything that touches the polygon
Constraints, the address, the property reference, the authority, the nation, the flood zone, listed buildings within 500 metres, planning history, energy certificates and comparable sales. Five parallel queries, each in its own error boundary.02
Traffic lights by category
Heritage, flooding, ecology, landscape and the rest, each resolved to red, amber or green from what was actually found rather than from a keyword in your description.03
A verdict, and a window
Two reds or more is high risk, and 16 to 26 weeks. One red is needs care. Two ambers is needs care. Anything else is likely approvable, at 8 to 12 weeks. The rule is that plain because a verdict you cannot reconstruct is a verdict you cannot argue with.04
Approval likelihood
A deterministic rules engine, not a black box. It starts at 86% and applies named coefficients, then folds in your council's own approval rate, the application type and the site area. The output is clamped between 5 and 95, because certainty in either direction would be dishonest.05
Your council, in detail
Approval rate, average decision time, Housing Delivery Test result, five-year housing land supply, standard-method housing need, levy rate and schedule, adopted plan and its year, and whether the tilted balance is engaged.06
Nearby decisions
The closest comparable applications from the decision record, with what happened to them.07
Documents and a price
The document set your route and your constraints require, itemised, with Wren's fee and your council's fee shown separately.
Then
Save it, and it keeps working.
A saved site re-appraises itself in the background, and can watch its own neighbourhood.
Set a radius and Wren emails you when something is filed near it. The alert keys on when we first saw an application, not on the date it was filed, and the difference is not academic: measured across the live table, the median gap between a filing date and the day it reaches us is 33 days. An alert built on filing dates almost never fires, which is exactly what the first version of this did.
Saved sites sit on a pipeline board, carry their own notes and colour, export as GeoJSON or CSV, and can be shared as a read-only link that shows the boundary, the constraints and the appraisal without giving anyone access to your account.
Start here
It takes about fifteen seconds.
Search a postcode, draw around your site, and read the answer. No account, no card.