ERZ
Eretzon
Denver, CO — modelled
Address in → feasibility package outDenver, CO — modelled

Address in. Feasibility package out — or an honest refusal.

A developer picks a market, types an address or parcel ID, and gets back the zoning rules that apply, the buildable envelope, a benchmarked cost range and the resulting numbers. A market is a way to browse and scope, not a unit of data: the address is resolved to its own local jurisdiction first — zoning in the US is a city and county function — and where that jurisdiction is not the market chosen, the platform renames both and answers with neither. Where the inputs to answer honestly are not in the database yet, the platform refuses and names every reason instead of returning a plausible-looking figure.

1 real parcel with a computed envelopeany address in Denver, CO answered live, plus 13 captured runs kept as the fallbackAcceptance harness: 9 of 9 at capture

Denver, CO — modelled — the reference market: district standards read out of the published Denver Zoning Code by hand, an ingested parcel register and zoning layer, and a cost band built from real City of Denver permit valuations

Berkeley, CA — reconnaissance only — no ordinance read, no zoning layer or parcel register ingested, no cost basis, and California's state layer unread Missing here: market_ordinance_unread, market_zoning_layer_not_ingested, market_parcel_layer_not_ingested, market_state_layer_unread, market_cost_basis_absent.

Outside the modelled markets — browse scope, not a data unit — an address anywhere the deployment does not model is refused by name, never priced with another market's numbers Missing here: market_not_modelled.

This field asks the engine about whatever you type — the engine runs on our own machine, this page's own server puts the question to it (your browser never reaches it, and nothing is computed in your browser), and every answer says where it came from: LIVE means the engine computed it when this page was built, CAPTURED means the engine could not be reached and the answer is one of the 13 real runs captured from that same engine on 10-07-2026 · 5:29 PM UTC at commit 1c90cb4, replayed verbatim. Under Denver, CO the captured inputs are 2323 curtis st · 2307 curtis st · 2307 curtis st mstr · 2323 curtis · 2307 curtis · 5800 tower road · 1100 congress ave, austin, tx 78701 · 0234219013000 · 0234219033999. Anything else gets an honest message rather than an invented package.

liveComputed by the engine just now, when this page was builtengine time 11-21-2026 · 9:29 PM UTCdataset denver-pilot-mixed · mixedEretzon Denver pilot dataset — sourced Denver source rows (parcel register, zoning polygons and hand-read zoning code, commercial building permits) plus a labelled FIXTURE set of modelling-assumption rows
POST /projects/lookup{"address":"2323 Curtis St","market":"denver-co"}HTTP 200status: completedata_quality: mixed
dataset denver-pilot-mixedcomputed 11-21-2026 · 9:29 PM UTC

Eretzon Denver pilot dataset — sourced Denver source rows (parcel register, zoning polygons and hand-read zoning code, commercial building permits) plus a labelled FIXTURE set of modelling-assumption rows

COST BASIS: the construction cost rate in this response is marked data_quality='derived'. It is our own derivation from the cited dataset named on the cost_baselines row in these records (see its source_name, source_url and source_note) — the source published the data underneath the figure, not the figure. It is an early screening range for advancing or killing a site: not a bid, not a budget, and not an engineering figure.

INPUT PROVENANCE (read off the records this response carries, not off a dataset label): of the 32 inputs that reached a figure, 2 are sourced (the parcel record; the zoning rule row); 5 are derived — our own arithmetic on a cited dataset (the cost baseline row; cost index row 29; cost.rate_basis.per_dwelling_unit; and 2 more); 25 are fixture, i.e. invented placeholders (cost.currency; cost.parking_type.baseline_reference_type; cost.soft_cost_mix.by_category; and 22 more). The 25 fixture rows are invented placeholders, and the figures they drive would move if those rows changed. The other 7 inputs are not fixtures — each one's own label is named above, and nothing in this response is invented beyond those 25 rows. Estimates are ranges, not guarantees.

Risk flags the engine raised (4)

Shown exactly as the response carries them — code, severity, message and evidence, nothing suppressed or softened.

  • fixture_datahigh25 assumption key(s)

    This package is built on FIXTURE data in part: 25 of the 32 inputs that reached a figure are labelled fixture — cost.currency; cost.parking_type.baseline_reference_type; cost.soft_cost_mix.by_category; and 22 more. Those 25 rows are invented placeholders, not sourced figures, and any figure they drive would move if they changed. No other input in this package is a placeholder: the provenance note in this response names what each of the others is. The arithmetic is real; those fixture rows are not sourced figures.

    rule and evidence
    {
      "rule": {
        "fires_when": "any input that reached a figure is labelled fixture",
        "fixture_input_count": 25
      },
      "evidence": {
        "data_qualities": [
          "derived",
          "fixture",
          "sourced"
        ],
        "fixture_inputs": [
          "cost.currency",
          "cost.parking_type.baseline_reference_type",
          "cost.soft_cost_mix.by_category",
          "cost.structure_height.band_upper_bounds",
          "envelope.avg_unit_size_sqft.by_building_type",
          "envelope.floor_to_floor_ft.by_building_type",
          "envelope.ground_floor_extra_height_ft",
          "envelope.net_to_gross_efficiency.by_building_type",
          "envelope.parking.max_surface_lot_coverage_pct",
          "envelope.parking.max_surface_spaces_per_acre",
          "envelope.parking.podium_max_stories",
          "envelope.parking.spaces_per_unit_default",
          "envelope.parking.sqft_per_space.by_type",
          "envelope.parking.surface_sqft_per_space_including_aisle",
          "envelope.parking.underground_min_stories",
          "envelope.residential_share_pct.by_building_type",
          "envelope.site.side_setback_yard_count",
          "feasibility.risk.high_inflation.threshold_pct",
          "feasibility.risk.high_inflation.window_quarters",
          "feasibility.risk.low_unit_count.threshold",
          "feasibility.risk.unusual_height.max_building_height_ft",
          "feasibility.risk.unusual_height.max_stories",
          "feasibility.risk.unusual_parking.max_parking_area_pct_of_gfa",
          "feasibility.risk.unusual_parking.max_spaces_per_unit",
          "feasibility.sensitivity.drivers"
        ],
        "input_count": 32
      },
      "data_quality": "fixture"
    }
  • unknown_geotechwarning

    No geotechnical study is recorded for this site. Foundation type, dewatering and soil remediation could move the cost materially and are not priced here.

    rule and evidence
    {
      "rule": {
        "geotech_known_is_true_required": true
      },
      "evidence": {
        "geotech_known": null,
        "parcel_id": "0234219013000"
      },
      "data_quality": "sourced"
    }
  • unknown_specification_levelwarning

    No finish specification level was given (budget / standard / premium). The cost baseline is priced at a mid-range specification, so a premium or budget scheme would move the numbers without any of the multipliers below changing.

    rule and evidence
    {
      "rule": {
        "accepted_values": [
          "budget",
          "standard",
          "premium"
        ]
      },
      "evidence": {
        "specification_level": null
      },
      "data_quality": null
    }
  • cost_not_escalated_to_run_quarterinfo

    The construction cost figure is in 2026Q3 money. The run is as of 2026Q4, 3 months later, and no escalation series we may apply covers the gap, so nothing was escalated: the figure keeps the vintage of the dataset it was derived from. It is not a today's-market figure, and carrying it forward is the reader's own step.

    rule and evidence
    {
      "rule": {
        "source": "cost_indexes.index_type = 'inflation_quarterly'",
        "applied_when": "no citable, displayable quarterly series covers the gap"
      },
      "evidence": {
        "baseline_quarter": "2026Q3",
        "run_quarter": "2026Q4",
        "months_not_covered": 3,
        "escalation_applied": 0
      },
      "data_quality": "derived"
    }

What resolved

The typed string, the parcel it resolved to, and the district whose rules were read.

Resolved parcel
0234219013000
2323 CURTIS ST · INDUSTRIAL-CHURCH · 9259 (INDUSTRIAL-CHURCH)
Address resolution
1 match
parcel_address_key — the address matched exactly one parcel
Site
15,800 sq ft
frontage 125.66 ft measured, not published by the register
Parcel ↔ polygon
agree
register says C-MX-5, the polygon says C-MX-5

What the code allows, not what stands. Every figure below is what C-MX-5 allows on this site under its rules — not a survey of what is on the site today. The register describes this parcel as INDUSTRIAL-CHURCH (prop_class 9259) with existing use INDUSTRIAL-CHURCH, and the engine carries that rather than assuming it away.

The buildable envelope

multifamily_5over2 on the GENERAL building form — footprint, height, floor area, units and parking, computed from the rule row for C-MX-5.

Dwelling units
78
binding: no_density_cap_configured · 850 sqft average unit at 84.00% net-to-gross
Gross floor area
78,929 sqft
of a 79,002 sqft maximum · FAR used 5.00
Storeys
5
zoning allows 5, height alone would allow 6
Building height
56.50 ft
10.50 ft floor-to-floor plus 4.00 ft of ground-floor height
Footprint
15,786 sqft
binding: program_floor_area · setbacks alone would allow 15,800 sqft
Frontage used
125.66 ft
buildable depth 125.74 ft — computed_equivalent_depth
Parking
86 spaces
1.10 per unit at a 1.10 ratio, per_dwelling_unit · assumption_default_zoning_silent
Parking structure
podium
rule_based_ladder: the 5-storey building is within the podium threshold (8), so parking goes in a podium rather than below ground · 28,380 sqft at 330 sqft per space

how the parking type was chosen (3 steps)
  1. surface capacity is min(15 spaces from 35% of the lot at 350 sq ft per space incl. aisle, 21 spaces from 60 spaces per acre) = 15
  2. required 86 spaces, open area left after the building is 14.29 sq ft
  3. surface cannot hold 86 spaces, so a structured option is needed; the building is 5 storeys and the podium threshold is 8

  • Envelope caveat. lot depth is not in the parcel record, so the envelope uses our own computed equivalent depth: lot area divided by measured frontage = 125.74 ft (method lot_area_over_measured_frontage). It is our method, not a code standard and not a surveyed depth: exact for a rectangular lot, and an upper bound on the buildable strip for an L-shaped, corner or through lot.
  • Envelope caveat. the zoning rule sets no floor area ratio; the envelope is therefore limited only by footprint and height, which overstates what is likely to be approvable.

The code states no maximum for max_far, max_lot_coverage_pct, min_parking_ratio — a positive statement about C-MX-5, not a missing value: no floor area ratio and no lot coverage maximum apply, so coverage does not bind here.

Protected District (§C): the 175 ft condition does not bite

C-MX-5 caps height within 175 ft of a Protected District at 75 ft, which binds the 95 ft incentive height rather than the 70 ft base. Whether that cap applies to this site is measured against the city's own polygons by the engine — never assumed, and never left to the reader to apply.

Verdict
does not bite
established true · applied true · bites false · refused false
Nearest Protected District
515.84 ft
U-RH-2.5 — shortest boundary-to-boundary distance from this parcel's own zone lot polygon to the nearest mapped protected polygon
Margin against 175 ft
340.84 ft
the measured distance is outside the threshold by this much · within tolerance of the threshold false
Measurement tolerance
0.0047 ft
computed from the geometries this measurement used, not chosen — round-off of the stored coordinates plus the projection's differential scale error. What it does not cover is stated in full below
Polygons the measurement read
133 measured
1 excluded as the site's own district polygon — a site cannot be its own protected neighbour
Protected polygons within 175 ft
0
none — nothing protected is inside the threshold
Unclassifiable polygons within 175 ft
0
none — every polygon inside the threshold was classified, so nothing was unreadable
Nearest polygon of any kind
D-AS at 0.00 ft
not_protected — 2010-code district, not named in DZC Art. 13 Div. 13.3's flat enumerated list (the definition has no character test)

The 175 ft line is ours, not the code's. the code fixes no line for the 175 ft: Article 13 was searched in full and the rule says what is capped ('all portions of a Structure … within 175 feet of a Protected District') without saying from or to what the distance runs. The line used here is OURS — the shortest horizontal distance from the parcel's zone lot polygon to the mapped protected-district polygon — and it can only over-trigger, which is the safe direction.

The nearest polygon of any kind is not_protected. D-AS at 0.00 ft — the engine's own flag says this site sits inside this polygon, and it is not the site's own district polygon (that is the one the measurement set aside above). A site can lie inside more than one polygon in the city's own layer, and a site inside a polygon is zero feet from it — the zero measures an overlap in the city's layer, not proximity: 2010-code district, not named in DZC Art. 13 Div. 13.3's flat enumerated list (the definition has no character test). A district the rule does not name cannot trigger it, however close it is.

The engine's own statement. the nearest protected district is U-RH-2.5 at 515.8 ft, outside the 175 ft threshold by 340.8 ft (measurement tolerance 0.005 ft), so the condition does not bite

the rule as printed, how the distance was measured, the tolerance in full, and the engine's own 3 documented limits

The rule the condition comes from. Denver Zoning Code §13.1.3.4.B (printed 13.1-13), 'Feet, within 175 feet of Protected District (max)': "Height of all portions of a Structure, developed with or without height incentives, that is within 175 feet of a Protected District shall not exceed the maximum height specified in the applicable building form table." For C-MX-5 that maximum is 75 ft (Article 7, §7.3.3, GENERAL building form table, printed 7.3-13), which binds the 95 ft incentive height, not the 70 ft base.

How we measured it. our conservative reading (the code defines no line for the 175 ft): the shortest horizontal boundary-to-boundary distance between the parcel's zone lot polygon and the mapped protected-district polygon, in a local equirectangular projection (lon x 111320 x cos(lat), lat x 110540, to metres; x 3.280839895 to feet). The code tests the structure and states the condition 'with or without height incentives'; measuring from the zone lot instead of the as-designed structure can only over-trigger, and a parcel that over-triggers is refused — the safe direction.

the search radius is a scope, not a threshold: the condition is decided at 175 ft, and a polygon within that distance of the site always has a bounding box within it of the site's, so the verdict does not depend on the radius

The tolerance, in the engine's own words. the tolerance is computed from the geometries this measurement used: round-off of the stored coordinates at the decimal places actually stored on the two polygons, plus the differential scale error of the local projection across the latitude span between them — 0.0047 ft for this pair. the tolerance covers the arithmetic of this measurement — the round-off of the coordinates as stored, and the differential scale error of the local projection over the latitude span actually measured. It does NOT cover the positional accuracy of the two publishers' geometries, which neither layer states; the measurement is as good as the two layers are, and that is why a distance inside the tolerance of the threshold is refused rather than answered.

  • the 'adjacent to Protected District' setback rows are subsumed by this test: anything adjacent to a protected district is within 175 ft of it, so they cannot bite on a parcel that passes the distance test. The nuance the code states — an intervening alley (public or private) preserves adjacency, an intervening named or numbered street destroys it — is not modelled: there is no alley right-of-way layer, and this is carried as a limitation rather than assumed away.
  • a height incentive (e.g. C-MX-5: 7 stories / 95 ft under Sec. 10.12.1) is recorded on the rule row and is never applied by this version: it is opt-in, and where the 175 ft cap applies the incentive path is removed — the cap binds the incentive height, not the base.
  • the tolerance covers the arithmetic of this measurement — the round-off of the coordinates as stored, and the differential scale error of the local projection over the latitude span actually measured. It does NOT cover the positional accuracy of the two publishers' geometries, which neither layer states; the measurement is as good as the two layers are, and that is why a distance inside the tolerance of the threshold is refused rather than answered.

the protected-district list, classified on the polygons themselves · Denver Zoning Code, June 25, 2010 | republished February 25, 2025
Denver Zoning Code, Article 13 (Rules of Measurement & Definitions), Division 13.3 (Definition of Words, Terms and Phrases), printed page 13.3-27 — 31 enumerated districts plus item 32; Former Chapter 59 §59-96(a) (printed p. 137) for the seven Former Chapter 59 'protected districts'
what was classified, when, and by whom · definition read by researcher (read 2026-09-29; evidence raw/protected-district-def-13.3-27.png)
zoning polygon layer vintage 2026-06-02; classified in our database on 2026-10-06; definition read: researcher (read 2026-09-29; evidence raw/protected-district-def-13.3-27.png)

The cost band

A range, never a single figure. Headline $152.00–$207.00 per sqft of gross floor area, midpoint $180.00 — in 2026Q3 money, basis 78,929 sqft.

Low
$12.00M
$11,997,143 at $152.00/sqft
Midpoint
$14.21M
$14,207,143 at $180.00/sqft
High
$16.34M
$16,338,214 at $207.00/sqft
Per dwelling unit — from the source
$158,315 – $216,338
the source-faithful form: the permits state a valuation and a unit count, not a $/sqft — this division is ours (mid $187,947)
Per dwelling unit — this run
$153,810 – $209,464
midpoint $182,143 over 78 units
Escalation
not applied
baseline 2026Q3 → run quarter 2026Q4: 3 month(s) uncovered, inflation factor 1.00, 0 series applied
Soft costs
100.00% hard cost
soft cost $0 — no land, no fees, no financing, no contingency: this is building work only
The band, from derived data
$152.00 – $207.00 / sqft
no confidence level is claimed — the band is the cost source's own low/high range carried through this run, not a statistical interval
Sensitivity around the midpoint
$12.01M – $16.40M
±15.44% observed spread in the sample (58 permits, 2023-02-01 to 2026-09-23) — a different question from the cost band above: geotechnical conditions and foundation type; finish specification level (budget / standard / premium); contractor market conditions at the time of bid; construction programme and carry cost
cost baseline 7 · derived · Denver
https://services1.arcgis.com/zdB7qR0BtYrg0Xpl/arcgis/rest/services/ODC_DEV_COMMERCIALCONSTPERMIT_P/FeatureServer/317/query
City and County of Denver — Commercial Construction Permits (daily extract from the Accela database) — 317, retrieved 2026-09-29
what the baseline covers, in its own words
The city's data states a declared construction valuation per permit and the dwelling-unit count on it. Dividing by the units on the permit, and then by the cited 1,044 square feet per unit to reach a $/sqft rate, are both our arithmetic, not the source's: the permits say nothing per square foot. The $/unit band is the source-faithful form of the same evidence and is displayed alongside the $/sqft headline, so a reader can see which of our two figures came from where. Across the 58 permits the middle half of $/unit runs $158,315-$216,338, which at 1,044 sqft per unit is $152-$207 a square foot (midpoint $187,947 -> $180); the full sample spans $122,856-$344,953 per unit, which is about $118-$330 a square foot, and that is why this is an order-of-magnitude screen to advance or kill a site and not a budget. The conversion rate is not ours: it is the U.S. Census Bureau's 2025 average of 1,044 sqft of gross floor area per rental unit, cited in unit_size_source on this row.
the rate this $/sqft figure is converted at: 1,044 sqft per unit
https://www.census.gov/construction/chars/xls/mfu_squarefeet_cust.xls
Characteristics of New Housing — Median and Average Square Feet of Floor Area in Multifamily Units Completed. average square feet of floor area per unit, units built for rent; 2025 (annual series; United States totals; new privately-owned multifamily units completed); read 10-06-2026. gross floor area per unit. The table's own footnote 1: 'The average was calculated using total square feet of all floors based on exterior dimensions divided by the number of housing units in the building.' That is the same basis as our $/unit to $/sqft conversion, which is why this series is the right one for it.

the baseline's full source note (3393 characters) and the engine's derivation

Construction cost is our own assumption, not market data. We price the building work at $152-$207 per square foot of gross floor area, using $180/sqft as the midpoint, in 2026 Q3 dollars: that is the vintage of the permit sample this rate comes from, and we do not carry it forward to the day you read it. It covers the building — structure, envelope, finishes and the parking podium — and excludes land, professional fees, permits, financing, contingency and construction escalation. The evidence is 58 City of Denver commercial building permits for new buildings of 40 or more units issued between February 2023 and September 2026. Each such permit states a declared construction valuation and the dwelling-unit count on it, and nothing per square foot, so two of our own arithmetic steps sit under this figure and both are shown. First, dividing the declared valuation by the units on the permit gives the source-faithful band of $158,315-$216,338 per dwelling unit (midpoint $187,947). Second, converting that to $/sqft at 1,044 square feet per unit gives the $152-$207 headline. The 1,044 is not ours: the U.S. Census Bureau's 2025 Characteristics of New Housing table on median and average square feet of floor area in multifamily units completed states 1,044 square feet as the average for units built for rent (https://www.census.gov/construction/chars/xls/mfu_squarefeet_cust.xls, read 6 October 2026), measured as total square feet of all floors by exterior dimensions divided by units — gross floor area per unit, the basis this conversion needs. No licence sentence was read on that source: the Census Bureau's policy pages returned 404, so we rely on 17 U.S.C. §105, which places works of the U.S. Government outside copyright. The band is not a confidence interval and carries no probability: it is the observed spread in the sample. Across the middle half of the 58 permits the $/unit figure runs 1.37x wide (P25-P75 $158,315-$216,338 around a median of $187,947); end to end the sample spans $122,856-$344,953 per unit, 2.81x, which at 1,044 square feet per unit is about $118-$330 per square foot. Treat this as an early screening figure, not a bid. Two cautions about what is inside the sample. The city's permit layer has no occupancy field, so the sample is not multifamily-only: where a lodging permit states UNITS they are guest rooms, and the hotels inside the sample pull the low end down. And the sample is not storey-verified: the layer states no storey count, so nothing in it confirms 5-over-2 podium construction specifically — the closest proxy we have, the 46 permits of 100-450 units, gives a nearly identical band. No regional, storey or escalation factor is applied on top: the sample is already Denver, already this building type, and as of September 2026 there is no free Denver construction-cost index to escalate it forward with. Soft costs are not added — we hold no sourced percentage for them, and a typical 25-35% of hard cost for fees, permits, financing and contingency is outside this figure. The split of the rate across site work, structure, parking, MEP and finishes is our own breakdown; the permits state only the total. Source: City and County of Denver, Commercial Construction Permits (daily extract from Accela, ArcGIS layer 317), retrieved 29 September 2026. Derived by us; not an official Denver figure, and NOT FOR ENGINEERING PURPOSES.

sensitivity.low/high = total_cost.medium x (1 -/+ band_pct), where band_pct is the observed spread in the sample the cost figure comes from. It is the observed spread in the sample the cost figure comes from (58 permits, 2023-02-01 to 2026-09-23): the middle half of it runs 158315-216338 per dwelling unit around a median of 187947, 1.37x wide, so the band is (p75 - p25) / (2 x median) = (216338 - 158315) / (2 x 187947) = 0.1544. Sample: City and County of Denver — Commercial Construction Permits (daily extract from the Accela database). This is a measured spread in a sample we hold, not a confidence level: no probability is attached to it, and it is distinct from total_cost.low/high, which is that same cost source's range carried through this run's multipliers. band_pct is stated to six decimal places and low/high are computed from that stated band, so these three figures agree exactly. The low/high band is the baseline's own range scaled by the same multipliers as the medium figure. It is a range over the source's own data, not a statistically derived interval. There is deliberately no confidence_level and no confidence_interval on this response: a range built from a sample this size supports no probability statement, so none is made and no nominal level is quoted. An earlier revision of this engine did quote one — 0.80 — and it was removed rather than replaced, because a claim with no evidence behind it is worse than no claim.

The zoning rules that produced it

Read by hand out of the published code and loaded with its citation — the engine resolves the row through one path with one effective date.

District
C-MX-5
Mixed Use 5 · jurisdiction Denver · rule row 7 · sourced
Height limit
70.00 ft / 5 storeys
the GENERAL form's own line
Minimum setbacks
F 0.00′ · S 0.00′ · R 0.00′
effective 06-25-2010
Overlays
none
an overlay changes which standards apply

Building form. These standards are C-MX-5's GENERAL building form — the form this row was read for. C-MX-5 states different numbers for its other building forms in the same tables, and a building developed under one of them is not governed by these values. The other forms, verbatim: TOWN HOUSE: Primary Street (min) 10 ft, Side Street (min) 7.5 ft, Side Interior (min) 5 ft. SHOPFRONT: height in feet (min/max) 24 ft / 70 ft, Primary Street / Side Street / Side Interior (min) 0 ft.

Cells the code leaves empty. The row names as genuinely stated absent in the code, so no maximum applies to them and none was invented.

Site-conditional and other-form rows (10): the other forms are carried verbatim, the Protected-District rows are computed

Each of these changes what the code means for a given site. The other-form and incentive rows travel with the answer rather than being averaged into it. The Protected-District rows above them are no longer only carried: whether the 175 ft cap applies to this site is computed from the city's own polygons, and the distance, the margin and the measurement's tolerance are in the Protected District block.

  • Feet, within 175’ of Protected District (max) = 75’ (C-MX-5, GENERAL form, printed page 7.3-13). Note (not code wording): computed, not carried — this cell is above C-MX-5’s own 70’ base maximum and the page does not explain it (C-MX-3 shows na). It is read as a site-condition variant on the height maximum, not as permission to build 5 ft taller. Whether it binds a given site is measured, not assumed: the engine takes the site’s own stored zone-lot polygon, measures the shortest distance to the nearest mapped Protected District polygon, and refuses the envelope where the 175 feet cap applies — the answer carries the protected_district block with the distance, the margin and the measurement’s own tolerance.
  • Sec. 13.1.3.4.B Feet, within 175 feet of Protected District (max), rule of measurement, verbatim: “Height of all portions of a Structure, developed with or without height incentives, that is within 175 feet of a Protected District shall not exceed the maximum height specified in the applicable building form table.” (printed page 13.1-13). Note (not code wording): the code does not state which line the 175 feet runs from or to.
  • Stories/Feet, with incentives (max) = 7/95’ (C-MX-5, GENERAL form), See Sec. 10.12.1, Height Incentives.
  • Side Interior, adjacent to Protected District (min) = 10’ (C-MX-5, GENERAL form).
  • Rear, adjacent to Protected District, where Alley (public or private) abuts a Rear Zone Lot Line (min) = 0’, and Rear, adjacent to Protected District, where Alley (public or private) does not abut a Rear Zone Lot Line (min) = 10’ (C-MX-5, GENERAL form).
  • Upper Story Setback Above 27’, adjacent to Protected District: Rear, alley/Rear, no alley and Side Interior (min) = 20’/25’, and Upper Story Setback Above 51’, adjacent to Protected District: Rear, alley/Rear, no alley and Side Interior (min) = 35’/40’ (C-MX-5, GENERAL form).
  • REQUIRED BUILD-TO — A Primary Street (min build-to % within min/max range) = 70%, 0’/10’, and Residential Only Buildings: 0’/15’ (C-MX-5, GENERAL form). Note (not code wording): this is a placement requirement, not a setback, and this row does not carry it.
  • PARKING (C-MX-5, GENERAL form, verbatim): Surface Parking between building and Primary Street/Side Street = Not Allowed/Not Allowed. Vehicle Access = Shall be determined as part of Site Development Plan Review. Note (not code wording): the table states no parking minimum.
  • Same district code, other building forms, verbatim from the same Division 7.3 tables: TOWN HOUSE (printed page 7.3-7) Primary Street (min) 10’, Side Street (min) 7.5’, Side Interior (min) 5’, Side Interior, adjacent to Protected District (min) 10’, Rear (min) 0’. SHOPFRONT (printed page 7.3-23) Feet (min/max) 24’/70’, Primary Street (min) 0’, Side Street (min) 0’, Side Interior (min) 0’. A building developed under one of those forms is not governed by this row’s values.
  • Protected District, the term the conditional rows above use, is defined in Article 13, Division 13.3 (printed page 13.3-27) as an enumerated list of zone districts S-SU-A through G-RH-3 plus any zone district retained from Former Chapter 59 and considered a protected Zone District under Section 59-96. Note (not code wording): it is a list, not a character test. The list is applied to our own zoning polygons — every ingested polygon’s district code is classified against it on the polygon row itself — and the condition is computed for each site from that site’s own polygon: the distance to the nearest protected polygon is measured, and the envelope is refused where the 175 feet cap applies and refused where the condition cannot be established.
zoning rule row 7 · sourced · ordinance null
https://www.denvergov.org/files/assets/public/community-planning-and-development/documents/zoning/denver-zoning-code/denver_zoning_code_article7_urban_center.pdf
Denver Zoning Code, Article 7 Urban Center, Section 7.3.3 Building Form Standards, GENERAL building form table (printed page 7.3-13), C-MX-5 column
coverage statement the engine attaches
the Denver zoning code sets no maximum lot coverage for C-MX-5: the applicable building form standards state no coverage maximum at all, so coverage does not bind this envelope. This is a positive statement about the code — the quantity does not exist — not a value that is missing or unread.

Every record that reached the answer is cited (5)

Table, row, what it decided, how good it is and where it came from — the citation chain the engine attaches to every response.

TableRowWhat it decidedQualitySource
parcels14767site geometry and site factssourcedCity and County of Denver — Parcels (ODC_PROP_PARCELS_A layer 245) · layer 245 · read 09-27-2026
zoning_rules7zoning limits stated for the GENERAL building form, and parking rule used by the envelope enginesourcedDenver Zoning Code, Article 7 Urban Center, Section 7.3.3 Building Form Standards, GENERAL building form table (printed page 7.3-13), C-MX-5 column
cost_baselines7hard cost rate, soft cost percentage and itemisation mixderivedCity and County of Denver — Commercial Construction Permits (daily extract from the Accela database) — 317, retrieved 2026-09-29
cost_indexes29city_cost adjustment (Denver)derivedCity and County of Denver — Commercial Construction Permits (daily extract from the Accela database) — 317, retrieved 2026-09-29
cost_indexes30structure_height adjustment (Denver|4-6)derivedCity and County of Denver — Commercial Construction Permits (daily extract from the Accela database) — 317, retrieved 2026-09-29

Dataset denver-pilot-mixed — Eretzon Denver pilot dataset — sourced Denver source rows (parcel register, zoning polygons and hand-read zoning code, commercial building permits) plus a labelled FIXTURE set of modelling-assumption rows

modelling assumptions used (27, of which 25 are still labelled fixture)
  • fixturecost.currency = USD — ISO currency code for every money figure.
  • fixturecost.parking_type.baseline_reference_type = podium — The parking type the parking_type index rows are expressed relative to (value 1.00).
  • derivedcost.rate_basis.per_dwelling_unit = [object Object] USD per dwelling unit — The source's own rate statement, out of which our $/sqft rate was converted: keyed '<city>|<building type>'. Shown beside the $/sqft figure so the conversion is visible.
  • fixturecost.soft_cost_mix.by_category = [object Object] share of soft cost — How each baseline's soft-cost percentage splits across categories. Must sum to 1.0.
  • fixturecost.structure_height.band_upper_bounds = [object Object] storeys — Story bands used to pick a structure/height index row. Ordered; the final null bound means 'and above'.
  • fixtureenvelope.avg_unit_size_sqft.by_building_type = [object Object] sqft — Average NET lettable unit size by building type; drives the unit count. These are OUR OWN programme assumptions — no published series states a net unit size — and they are marked fixture. Note the basis, because it was the source of a real error: this is net area, so it is NOT the divisor for a $/unit-to-$/sqft conversion, which needs gross floor area per unit. For multifamily_5over2 the gross equivalent is 850 / 0.84 = 1,012 sqft per unit, within 3% of the U.S. Census Bureau's 2025 average of 1,044 sqft of gross floor area per rental unit (the figure the cost conversion now uses — see the unit_size_source block of the Denver cost basis file). Reading this net value as a gross one is what made the cost headline read $185-$255/sqft. Changing any value here changes the unit count and the envelope.
  • fixtureenvelope.floor_to_floor_ft.by_building_type = [object Object] ft — Storey-to-storey height by building type, used to turn a height limit into a storey count.
  • fixtureenvelope.ground_floor_extra_height_ft = 4 ft — Extra clear height on the ground floor (lobby or retail) added to the building height.
  • fixtureenvelope.net_to_gross_efficiency.by_building_type = [object Object] ratio — Net-to-gross efficiency by building type; drives the unit count.
  • fixtureenvelope.parking.max_surface_lot_coverage_pct = 35 % — Most of the lot that surface parking may cover; used only for the parking-type ladder.
  • fixtureenvelope.parking.max_surface_spaces_per_acre = 60 spaces/acre — Practical ceiling on surface parking density; used only for the parking-type ladder.
  • fixtureenvelope.parking.podium_max_stories = 8 storeys — Above this storey count the ladder stops choosing podium parking.
  • fixtureenvelope.parking.spaces_per_unit_default = 1.1 spaces/unit — Parking ratio used ONLY when the zoning rule states none. Using it is recorded as an assumption, never presented as a code requirement.
  • fixtureenvelope.parking.sqft_per_space.by_type = [object Object] sqft/space — Built area per parking space including aisles and circulation.
  • fixtureenvelope.parking.surface_sqft_per_space_including_aisle = 350 sqft/space — Land area per surface space including drive aisles; used only for the parking-type ladder.
  • fixtureenvelope.parking.underground_min_stories = 9 storeys — At or above this storey count the ladder goes underground rather than podium.
  • fixtureenvelope.residential_share_pct.by_building_type = [object Object] % — Share of gross floor area that is residential, by building type.
  • fixtureenvelope.site.side_setback_yard_count = 2 yards — How many side yards the side setback applies to.
  • fixturefeasibility.risk.high_inflation.threshold_pct = 6 % — Cumulative construction inflation over that window that triggers the high-inflation flag.
  • fixturefeasibility.risk.high_inflation.window_quarters = 8 quarters — How many recent quarters the high-inflation flag looks back over.
  • fixturefeasibility.risk.low_unit_count.threshold = 10 units — Below this unit count the per-unit figures are flagged as volatile.
  • fixturefeasibility.risk.unusual_height.max_building_height_ft = 150 ft — Above this height the building is flagged as unusually tall for this model.
  • fixturefeasibility.risk.unusual_height.max_stories = 12 storeys — Above this storey count the building is flagged as unusually tall for this model.
  • fixturefeasibility.risk.unusual_parking.max_parking_area_pct_of_gfa = 40 % — Above this share of gross floor area the parking provision is flagged as unusual.
  • fixturefeasibility.risk.unusual_parking.max_spaces_per_unit = 1.5 spaces/unit — Above this ratio the parking provision is flagged as unusual.
  • derivedfeasibility.sensitivity.band_pct = [object Object] ratio (half-width of the sample's interquartile range, share of its median) — OBSERVED SPREAD IN THE SAMPLE — the sensitivity band. It is a fact about the cost sample (58 permits, 2023-02-01 to 2026-09-23), not a chosen band, not a programme assumption and not a confidence level: (p75 - p25) / (2 x median) = (216338 - 158315) / (2 x 187947) = 0.1544. Half the observed interquartile range as a share of the median, so the band reproduces the spread the sample actually has. Where a market has no entry, no band is invented for it.
  • fixturefeasibility.sensitivity.drivers = geotechnical conditions and foundation type,finish specification level (budget / standard / premium),contractor market conditions at the time of bid,construction programme and carry cost — The programme factors the sensitivity band is intended to cover.

engine_versions {"zoning":"0.1.0","envelope":"0.1.0","cost":"0.1.0","feasibility":"0.1.0","orchestrator":"0.1.0"} · run took 550 ms · project_id · computed live by the engine when this page was built, against the deployment's own database

Denver, CO — modelled — the reference market: district standards read out of the published Denver Zoning Code by hand, an ingested parcel register and zoning layer, and a cost band built from real City of Denver permit valuations. The engine confirmed the scope: this address resolved to Denver, the market denver-co. market=denver-co travels in the URL with the address.

Compare with 2307 Curtis St →The same angle, the other Curtis Street address: both in C-MX-5 on CURTIS ST, one returning a full envelope and the other a refusal with its reasons named.

Every figure, reason code, row id and citation below was returned by the running API or derived from the live database — this page does not type its numbers by hand. A LIVE answer was computed by the engine when this page was built; a CAPTURED one was captured from that same engine on 10-07-2026 · 1:26 PM UTC at commit 409ad7a, and every answer says which of the two it is. Three things here are written by us rather than generated, and we would rather name them than let the sentence above cover them: the acceptance count in the pill (nine checks in scripts/acceptance_denver.py, which the snapshot generator refuses to write past if any of them fails), the count of district codes holding no relief-free parcel, and the prose — including “What is not built yet”, which is a list of what we have not done. The market list in the dropdown is not written by us at all: it is the engine's own GET /markets response, captured with the runs, so the coverage claim on the page — one market modelled, one in reconnaissance — is the engine's claim, not our wording.

That one envelope parcel is 2323 Curtis St — a single register record in C-MX-5. Its figures are what the district's rules allow on the site, not a survey of what is there today: the note directly above the numbers says what the register records on it.

What comes back
envelope · cost · feasibility
footprint, height, floor area, unit count, parking, then a cost range — all from the rules that actually apply to the parcel
When we cannot answer
refuse, and name why
a missing standard, an unmeasured frontage or an unresolved overlay stops the number rather than guessing it
Every record is cited
source · vintage · read date
no figure reaches a response without the row it came from, labelled by how good that row is
The flow that works today

One real address, run end to end

This is the acceptance case: the owner's own address, run through the real pipeline and checked by an automated harness. It is deliberately the hard case — a pre-2010 district with an overlay, waivers, a use-form sentinel and no published frontage — so the refusal is the feature, not a failure.

POST /projects{"address": "5800 Tower Road, Denver, CO 80249"}status: manual_review_requiredHTTP 201
Resolved parcel
0010208160000
5800 N TOWER RD · RESIDENTIAL-MULTI UNIT APTS · 120 units existing
Address resolution
152 candidates
parcel_address_key; the master parcel was selected from the unit records and the choice is reported back
District
C-MU-20
Former Chapter 59 Zone · code vintage 1956 · ordinance 131/2000
Site
264,461 sq ft
frontage: no value — this parcel is outside the C-MX-5 measurement scope
Overlay
AIO
an overlay changes which standards apply
Waivers / conditions
WVRS · *
site-specific relief granted by ordinance
Use-form code
999
999 is the sentinel for ‘not a 2010 building form’
Parcel ↔ polygon
agree
register says C-MU-20, polygon says C-MU-20

Refused for 7 named reasons

  • rules_not_transcribedhigh

    the numeric standards for this district (setbacks, lot coverage, FAR, height in feet, minimum parking) have not been transcribed into zoning_rules yet, and no US city publishes them as data. No limit is invented here: the district identity below is what the data supports.

  • former_chapter_59_districthigh

    this is a Former Chapter 59 (pre-2010) district. It sits outside the 2010 Denver Zoning Code building-form standards, so the modern standards do not govern the site — applying them would be confidently wrong.

  • overlay_districthigh

    the parcel is inside an overlay district. An overlay changes the standards that apply, and its provisions are not modelled, so the base-district numbers (even once transcribed) would not be the whole answer.

  • waivers_or_conditionshigh

    the zoning layer records waivers and/or conditions against this site. Site-specific relief like this is granted by ordinance and must be read before any standard is applied.

  • zone_use_form_sentinelhigh

    the zoning layer gives this site the use-form code '999', its sentinel for 'not a 2010 building form'. A building-form standard cannot be selected from it.

  • lot_frontage_unknownhigh

    the parcel register does not publish a street frontage, and the front setback cannot be applied without one. Supply the frontage (or a surveyed site plan) rather than accept a footprint derived from a guess.

  • cost_baseline_not_sourcedwarning

    the construction cost basis that priced this response is ours, not market data: it is labelled 'derived' and we derived it from City and County of Denver — Commercial Construction Permits (daily extract from the Accela database) — 317, retrieved 2026-09-29. The city published the data underneath the figure; nobody published the figure. Its limits and its citation travel with it on this response (cost.baseline, and the cost_baselines record in the provenance), and it is a screening range for advancing or killing a site — not a bid, a budget or an engineering figure. Reported rather than refused because the basis is specified, cited and labelled: the envelope and the money may be computed, and the flag stays visible next to them.

Nothing numeric was returned

envelope
null
cost
null
feasibility
null
sensitivity
null

The cost basis for this run is our own derivation from a cited dataset, not a published figure. It is reported as a flag rather than a refusal: the basis is specified, cited and labelled, and the label travels with the number.

The exact numbers still missing for this district

Read back from the response — the standard must supply all of them, or the district gets no row at all:

max_dwelling_units_per_acremax_farmax_height_ftmax_lot_coverage_pctmax_storiesmin_front_setback_ftmin_lot_area_sqftmin_parking_ratiomin_rear_setback_ftmin_side_setback_ft

Every record that reached the answer is cited

TableRowRecordQualitySource
parcels83330010208160000 — 5800 N TOWER RDsourcedCity and County of Denver — Parcels (ODC_PROP_PARCELS_A layer 245) · layer 245 · read 09-27-2026
zoning_districts3626C-MU-20 — Former Chapter 59 ZonesourcedCity and County of Denver — Zoning (CCD_Zoning layer 27) · layer 27 · read 06-02-2026

INPUT PROVENANCE (read off the records this response carries, not off a dataset label): of the 2 inputs that reached a figure, 2 are sourced (the parcel record; zoning_districts row 3626). No input in this response is a placeholder: nothing here is invented, and every labelled input above is either sourced or derived from a cited dataset. Estimates are ranges, not guarantees.

What the database already holds

Real counts from the primary pilot scope

These are the numbers that decide which district can produce a first answer, and they are generated by a script against the live database rather than typed into a slide.

12,407
parcels in the pilot scope
all with stored boundary geometry
92%
on the current 2010 code
11,418 on the 2010 code; of the rest, 988 are Former Chapter 59 and one joins to no zoning polygon
6,560
relief-free parcels
no overlay, waiver, PUD or former-code flag
180
parcel ↔ polygon disagreements
surfaced and refused, not hidden — 1.45% of the scope
District codes present
66
of which 41 hold no relief-free parcel at all and can never produce an envelope
Former Chapter 59
988
pre-2010 ground we deliberately never transcribe
Joined to a polygon
12,406 of 12,407
1 joined to none, 3 with nothing checkable
Scope
central mixed-use corridor
bbox -104.99, 39.74, -104.955, 39.77 — the densest current-code mixed-use ground in the city

Why parcels are held back from an answer

A parcel carrying any of these is refused rather than modelled, because each one changes what the code means for that site.

  • 988former_chapter_59_district
  • 4,391overlay_district
  • 218planned_unit_development
  • 737waivers_or_conditions
  • 1,866zone_use_form_sentinel

Street frontage: measured for C-MX-5, with two things we will not smooth over

The Denver parcel register publishes no frontage field of any kind, so every frontage in the database is ours: the street-facing lot edge, measured against the city's own street centrelines, with the source, the tolerance and the method labelled on every single value. For C-MX-5, 395 of 455 parcels now carry one; 60 carry none and a named reason (no_parallel_street_edge), because a zero would be a wrong number wearing a number's clothes. Range so far: 2–465.49 ft, median 122.42 ft.

  • Some "parcels" are not sites. The smallest readings are honest geometry on land nobody can build on: 0227129055000 (3721 N MARION ST APPRX) is 2 ft; 0235514036000 (2200 E 18TH AVE APPRX) is 2 ft; 0234921021000 (605 E 16TH AVE) is 15.4 ft. The edge genuinely exists and genuinely faces the street, so the measurement is right and the site is wrong. We did not invent a minimum plausibility bound to hide them — that would mean retuning a tolerance we derived from the data, and every number above would have to be re-measured.
  • Many "lots" are stacks. 178 of 455 parcels share their stored boundary with at least one other parcel — 292 distinct boundaries among them, the largest group being 62 parcels on one boundary. These are condominium and unit stacks, so each unit inherits the parent footprint's frontage: honest as geometry, misleading as "the lot". Both this and the strips above now refuse with their own named reason — boundary_shared_with_other_parcels and recorded_boundary_not_a_site — rather than being modelled from a frontage that measures the parent footprint. The 2307 Curtis St run on this page is one of them.

Measurement is C-MX-5-only so far: 395 of 12,407 corridor parcels and 0 of 9,141 in the airport scope. The rest still refuse with lot_frontage_unknown — which is why the acceptance case above still refuses. 5 frontage values in the database are still fixture test rows, not measurements, and are labelled as such.

The district the first numbers come from

C-MX-5, the district those runs came out of

C-MX-5, in the central corridor: the district the first numbers come from, spread across the district rather than concentrated on one block. It is not the largest code in scope — C-MX-12 holds 456 parcels to C-MX-5's 455 — but all 456 of C-MX-12's carry an overlay and none of them is relief-free, while 204 of C-MX-5's are.

455
C-MX-5 parcels
204 of them relief-free
29
city zoning polygons
the number of C-MX-5 polygons in scope — a polygon count, not a claim that the table is uniform: the same district still carries per-form numbers and overlays
3
register disagreements
out of 455 — each one refused with both codes named
22 + 78
further relief-free parcels
each a SEPARATE district needing its own read — the C-MX-5 row names C-MX-3 and G-MX-3 as not covered, so these are reachable work, not a free extension
If that table does not read clean
U-MX-3
122 relief-free of 135 parcels, with zero register disagreements
Rule for every district
no row unless every value is read
each required value must be either a single unconditional number, or one the code positively states none of — the engine records that second case as stated_absent and says so in the answer. A half-read district gets no row, because a guess would turn a refusal into a number
Honest status

What is not built yet

Listed plainly, because the value of this page is that it does not overstate where the product is.

PieceStateWhat it is waiting on
Address → parcel → district → refusalworksNothing. This is the flow shown above, checked by the harness on every run.
Zoning rules for C-MX-5worksThe GENERAL building-form row was read out of the published code by hand and is loaded live with its citation, effective date and the cells the code states absent. It prices every run the search box answers. C-MX-5's other building forms are carried verbatim rather than averaged into the answer. The site-conditional Protected District rows are computed, not carried: every ingested zoning polygon is classified against the code's own enumerated list, the distance from the site's own zone lot polygon to the nearest protected polygon is measured to that measurement's own stated tolerance, and the parcel is refused where the 175 ft cap bites — or where the condition cannot be established.
Street frontagein progressMeasured for C-MX-5 only: 395 of 455 parcels, 60 with a named reason and no value. The rest of the corridor still has none. The two data-quality cases above no longer publish a frontage at all — they refuse with their own named reasons, boundary_shared_with_other_parcels and recorded_boundary_not_a_site.
Cost benchmarkin progressA derived band read from real City of Denver commercial construction permits, labelled 'derived' and carried with its citation on every response. The modelling assumptions underneath it — average unit size, net-to-gross efficiency, storey height, parking areas — are still fixture rows, which is why the cost-bearing packages above are labelled 'mixed' and carry a high-severity fixture flag.
Buildable envelopeworksComputed for C-MX-5 today: units, floor area, storeys, height, footprint, frontage and parking all come back in the runs above, each with the constraint that bound it and every caveat attached.
Feasibility numbers + risk flagsworksThe cost band, the $/unit figures, the sensitivity band and the risk flags all come back with the envelope. Missing limits are reported as data flags, never filled in — and the high-severity fixture flag stays on the package until those assumption rows are sourced.
Marketsin progress3 markets are listed and all but one refuse by name. Denver, CO is modelled — the reference market, the only one with transcribed district standards, ingested layers and a cost basis. Berkeley, CA: reconnaissance only — no ordinance read, no zoning layer or parcel register ingested, no cost basis, and California's state layer unread; Outside the modelled markets: browse scope, not a data unit — an address anywhere the deployment does not model is refused by name, never priced with another market's numbers. The list itself is the engine's own GET /markets response, captured with the runs, so a market can only be described as the engine describes it.
Search in the browserin progressThe market dropdown and the field at the top answer a fixed set of 13 runs produced by the real engine and captured verbatim, each filed under the market it was asked in, and the page says so in the answer itself. A market whose data this deployment does not hold cannot be answered at all — the engine refuses it by name, and the page shows that refusal rather than a package borrowed from Denver. Answering an arbitrary address needs the API hosted, which we are not paying for yet.
PDF report, share link, accounts, saved projectsnot startedAfter the first computed envelope.