Methodology

How every verdict is built.

A public methodology for the same reason we cite our sources: research is only as trustworthy as the process behind it.

Research to verdict

From raw data to a clear decision.

Four stages. Each one cites what we know, what we do not know, and how confident we are. Click any step to read the full method.

Step 01
Collect
AI research agents gather project brochures, RERA filings, municipal records, court notices, market indices and builder dossiers.
Evidence
Every data point is tagged to a primary source and a retrieval date.
Limitation
Some municipal records are published with delays of several weeks.
Step 02
Verify
Claims are cross checked against original documents. Contradictions between a builder statement and a public filing are flagged, not ignored.
Evidence
RERA registration numbers, land use maps and litigant names are matched before inclusion.
Limitation
Verdicts reflect the latest available public record, not hidden private agreements.
Step 03
Synthesize
Findings across the eight pillars are weighted and reconciled into a single confidence range with reasoning written in plain language.
Evidence
Each pillar score is derived from multiple independent signals, never from one source alone.
Limitation
Synthesis is probabilistic; edge cases may require an analyst to review manually.
Step 04
Decide
You receive a clear verdict, cited evidence, confidence range and the key risks to consider before signing a cheque.
Evidence
Final reports include a source list, timestamp and the specific questions the research answered.
Limitation
A verdict is research, not legal or investment advice. Always consult a qualified professional.

Research methodology

Draft with AI. Verify with sources. Review with humans.

Our methodology is public because our business depends on it being defensible.

Collect

AI helps draft the research from public sources such as RERA filings, municipal records and court records. Findings are checked against those sources, and Pro reports are reviewed by an analyst.

Verify

Findings across the eight pillars are reconciled against one another. Contradictions, such as a builder claim vs an RERA filing, are flagged, not smoothed over.

III

Synthesize

Signals from every pillar are weighted and reconciled into a single confidence range. No pillar is allowed to override the others without a documented reason.

Decide

A human analyst reviews confidence flags, legal amendments and edge cases. The final verdict carries a calibrated confidence range, not a false precision number.

Research notes

What the research labels represent.

The homepage dossier is an illustrative example, not a completed property assessment. These notes describe the evidence categories, not verified findings about a specific property.

RERA Evidence

Review the relevant state RERA portal for project registration, declared timelines, filings and disclosed updates. Match the registration to the project and record the source and retrieval date. Registration alone does not establish title, construction quality or completion.

Builder Review

Compare documented delivery history, project disclosures and relevant public regulatory or court records. Distinguish allegations from resolved findings and check entity identity. Past performance is context, not a guarantee of future delivery.

Price Analysis

Compare available transaction records and published market data for similar properties in the same locality. Separate asking prices from recorded transactions and account for size, condition and additional charges. Sparse or dated data limits confidence; estimates are not valuations or return forecasts.

Location Risk

Review municipal plans, official flood or environmental records where available, and current transport and infrastructure information. Separate existing facilities from proposed works and verify property level access. Locality data may not capture conditions at an individual site.