Filing & company facts
Filing identifiers, classifications, carrier relationships, lifecycle dates, reported rate impact, program premium and affected policyholders where stated.
Company-level values stay aggregated at the company level.
RATEFILEAI DATA PRODUCTS
The filing record.
The filing record.
Ready for your analysis.
Structured P&C filing facts and carrier analytics for teams that want to work in their own research tools, models and data environments.
Part of RateFileAI. A data delivery offering—not another application subscription.
ABOUT DATA PRODUCTSSpend less time assembling records.
Spend less time assembling records.
More time answering the question.
RateFileAI Data Products turns public filing observations into normalized, joinable data. Use filing and company-level facts alongside documented calculations to study carrier activity, reported changes and regulatory review.
Built for product, pricing, actuarial, regulatory, competitive-intelligence and data teams. Use the RateFileAI application to explore filings, or discuss a bounded data package for analysis in your own environment.
WHAT YOU RECEIVE
Data you can join. Calculations you can inspect.
The prepared pilot pairs structured facts with definitions, measurement boundaries and reproducible examples.
Carrier analytics
Filing counts and mix, cadence, review timing, pending age, explicit approval outcomes, impact distributions and comparable-peer views.
Calculations retain observation counts and exclusions.
Release documentation
Typed Parquet tables, a data dictionary, field availability, methodology, limitations, a versioned manifest, checksums and 20 SQL examples.
Evaluation samples and delivery require an agreed scope.
The package contains facts and calculations—not PDFs, ZIP archives, policy wording, raw correspondence or personal contact details. It does not grant application, API or source-document access.
MEASURED PILOT COVERAGE
A defined slice, not a blanket promise.
557 Texas Personal Auto filings, submitted Sep 1, 2024–Sep 21, 2026. Counts below describe the prepared Sep 22 copy, not all Texas filings or the full five-year dataset.
| Fact family | Observed | How to read it |
|---|---|---|
| Carrier relationship | 552 of 557 filings | 118 distinct writing carriers; a multi-company filing can relate to more than one carrier. |
| Reported overall rate impact | 438 of 557 filings | A carrier-stated filing value, not an independently confirmed approved rate. |
| Reported premium change / affected policyholders | 437 / 438 filings | Signed premium change and reported policyholder count are separate measures; people are not deduplicated across filings. |
| Form schedule facts | 215 observations · 83 filings | Form numbers, titles, types, editions and stated actions—not full policy wording. |
| Rule schedule items | 435 observations · 158 filings | Schedule references and actions. A replacement entry does not by itself prove a numerical rule change. |
| Dated review events | 1,820 observations · 493 filings | Explicitly dated objections, responses and supported dispositions; not a complete correspondence history. |
Know the limits before using the data.
Non-rate and actuarial extraction is selective. Coverage-change facts, compatible exposure bases and requested-versus-approved rate comparisons are not yet populated in this pilot. Unprocessed attachments and unreadable pages remain; an empty field does not mean a change never occurred.
Basis: schema-v5 private pilot, 557 accepted filing records; base observations as of Sep 22, 2026 at 17:37 UTC, held-source enrichment at 20:24 UTC. Availability is not extraction accuracy. Other states, lines, periods or update schedules require a separately confirmed scope.
METHODOLOGY
Keep the source, the fact and the calculation distinct.
Each release has a defined population and observation date. Unknown, not reported and not applicable are never silently changed to zero.
Define the population
Bound the jurisdiction, line and submission period. Preserve stable filing and carrier identities; disclose acquisition and field-coverage gaps.
Extract supported facts
Read explicit source fields, schedules and dated events. Normalize units and classifications while retaining source provenance and the difference between stated and calculated values.
Calculate with context
Keep denominators, eligible dates and exclusions alongside metrics. Separate explicit approvals from broader effective outcomes; do not label open review age as approval time.
Validate the release
Check schemas, joins, units, content boundaries and reproducibility. Report source-comparison work separately: technical checks alone do not establish population-wide extraction accuracy.
Reported premium and policyholder totals are filing observations, not a deduplicated insurance book. No market share, profitability, unique-customer count or carrier-quality rating is inferred.
PRICING & SCOPE
Start with the data your work needs.
Discuss the states, lines, time period, fields and intended use with RateFileAI. Coverage, delivery and pricing are confirmed for the agreed scope.
No published Data Products price or self-service checkout is active. A pricing inquiry is not an order, a subscription or immediate access. The current prepared pilot is a fixed copy; recurring feeds and service levels are not included by default.
Scope-led inquiryRequest pricing
Include “Data Products” and your required market, period and use case in your message.
Compare with the application planA few useful distinctions.
Is this separate from RateFileAI?
No. Data Products is part of RateFileAI. It supports analysis in your own tools; the application supports interactive Search, filing review and monitoring. Data delivery rights are agreed separately from an application subscription.
Can I download the underlying filing archive?
No. This offering is a structured facts-and-analytics package, not a public archive of source documents. The authenticated application's source-document access follows its own permissions and delivery rules.
Does the pilot include every filing or change?
No. The stated pilot is a bounded observed population. A represented carrier, a supported field or a processed document does not establish complete market or attachment coverage. Every agreed delivery must retain those limitations.
Does an approval rate mean the proposed change was approved unchanged?
No. Explicit approval metrics use a stated decision denominator and remain separate from effective-disposition metrics. Neither establishes an approved-versus-requested rate comparison or an implemented customer price.