Ask better questions
The answers are buried across maps, regulations, market data, assumptions and financial models. Civexa brings them together — and shows the evidence behind each answer rather than asserting it.
Why is it so hard to get a straight answer about a piece of land?
Capacity engine · zoning from the adopted code, developable area from the screened site layers. Every figure links to the rule or the layer that produced it.
Site engine · constraints are unioned rather than summed, so shared ground is not counted twice. An unscreened layer is reported as a gap, never as a clear result.
King County GIS recorded-sales layer. These are recorded transaction prices, not listings or estimates.
HUD Small Area Fair Market Rents for this ZIP. These are programme benchmarks — 40th-percentile gross rent including utilities — not market asking rents, and Civexa labels them that way rather than presenting them as comparables.
Finance engine · hard cost from the construction-cost band for this product type, every other line computed from it with its formula shown. The band is flagged as the assumption most worth replacing with a contractor's quote.
Residual land value is what this project can support at this target return under these assumptions. It is explicitly not an appraisal and not a negotiating position — change the target and it changes with it.
Illustrative answers from a real Civexa run. Every figure is one the engines computed, with its source named beneath it.
The model reads the answer; it does not compose one
Civexa's AI calls the structured result — the same ledger of metrics, formulas and inputs the dashboard renders — and explains it. It does not perform arithmetic in prose, and it does not paraphrase a number into existence. An answer names the scenario it belongs to and the material assumptions behind it.
A figure without provenance is an opinion
Each value in an answer keeps the label it had in the model: read from GIS, derived from the code, or a screening estimate awaiting your input. The same labels appear in the dashboard and in the exported report, so the answer and the document agree.
Which is the answer that matters most
Where the underlying model could not establish a value, the answer says so and names what is missing. That is more useful than a confident sentence built on an assumption nobody stated.
This is the same discipline the engines follow: an unscreened layer is a gap, not a clear result, and a missing price means no return rather than an assumed one.
Ask Civexa AI
Does the AI make up numbers?
Can it answer about any property?
What can I ask?
Every property has a possibility. The question is whether you can see it.
Civexa AI helps you find out.