01
Acquire without mutation
Pull Seattle’s latest 90 days of permit records while retaining an untouched raw layer for auditability and reruns.
Seattle permit intelligence
Test the signal before building the product.
BuildSignal turns 90 days of public permit records into an auditable prototype for evaluating commercial opportunities—without hiding source quality or overstating AI confidence.
A focused validation engagement—not a dashboard, subscription, or marketing system.
What to expect
Evidence over theatre.
Raw stays raw
Original records remain unchanged and traceable.
Rules stay repeatable
Core calculations are deterministic and reviewable.
Inference stays labeled
AI output is separated, explained, and confidence-scored.
The prototype
One reproducible path from permit record to qualified signal.
Each stage produces an inspectable artifact, so stakeholders can decide whether the source is valuable enough to automate or expand.
02
Measure what the source can support
Profile completeness, dates, addresses, values, statuses, permit types, and commercial or residential signals before relying on them.
03
Classify opportunity context
Use deterministic rules for cleaning and scoring, with AI reserved for descriptions, project scope, and likely downstream needs.
04
Review the strongest examples
Inspect 25 high-value candidates with source fields kept separate from inferred fields and confidence attached to every inference.
05
Document limits before scaling
Surface missing data, ambiguous records, automation risks, and the next tests required before operational use.
06
Stay portable by design
Keep acquisition and transformation boundaries clear so another city’s Socrata or open-data source can replace Seattle later.
Engagement
A fixed-scope validation pilot.
Pricing is scoped after a short source and stakeholder review, because record volume, access constraints, and review depth affect the work.
Seattle 90-day pilot
Custom fixed fee
- Preserved raw source export
- Cleaned 90-day working dataset
- Field completeness and quality report
- Permit and opportunity classification
- Deterministic scoring proposal
- 25 reviewed opportunity examples
- Source and AI fields separated
- Risk and next-step recommendations
Built for a go / no-go decision
The final recommendation addresses reliability, useful segments, automation risks, portability, and whether further investment is justified.
Details
Clear boundaries, before work begins.
Start with the evidence
Find out if Seattle permits contain a usable signal.
Define the decision, inspect the source, and scope a prototype that can be audited before it is scaled.