Every figure traces back to a named public system — OpenDataPhilly payroll, the Office of Property Assessment, Licenses & Inspections, 311, public court filings. Nothing is estimated or generated without a source. The full list is on Data & Sources.
Update frequency varies by source — some feeds refresh continuously, others quarterly or as-published. ALKARTIS is only as current and accurate as the source agency's own data, and any error present in the original public record can carry through to what's shown. That's stated plainly, not buried in fine print.
Where a product makes a recommendation — whether an assessment is worth appealing, whether a pattern across properties is significant — the comparison behind it is shown, not just the conclusion. And the honest negative is treated as a real outcome: not every assessment should be appealed, and not every landlord with one violation is a pattern. Saying so is part of the product, not a caveat bolted on afterward.
ALKARTIS connects public records that are already public by law — it doesn't introduce new personal data into the public domain. See the Privacy Policy and Terms of Service for the current, general-purpose versions of both, ahead of formal entity formation.
If a figure looks wrong, that's worth knowing about directly — reach out and it gets checked against the source record.
Where AI is part of a product — summarizing, classifying, or explaining already-connected records — it operates on grounded, sourced data rather than generating claims from scratch. See Why AI Isn't Enough for the fuller argument.
This is the same standard the rest of the section holds ALKARTIS to. See it applied to a real question.