What we're building next for the empirical record, in the order we expect to ship it. Every item names the dependency it's waiting on — when the dependency lands, the feature follows.
Leave an email on anything below and we'll write once — when it launches.
Work that is actively running against the corpus today.
In progress — the classifier is running over the citation record now.
Every citation between cases, classified: did the later court follow the earlier one, explain it, distinguish it, criticize it, or overrule it? Each label shows the exact sentence it rests on, with an accuracy sample published alongside the methodology.
In progress — search infrastructure provisioning is underway.
Full-text search across the entire opinion corpus, ranked with citation awareness — and a semantic layer that finds passages by what they mean, kept only if it measurably beats keyword search.
Unblocked and sequenced — each builds on the verified quotation record.
Unblocked — builds on the verified quotation record.
For each section of the U.S. Code and the state codes we host: the decisions that applied it, grouped by topic where the volume warrants, each note grounded in the citing sentence — the annotated-statute experience, built from the record instead of an editorial staff.
Unblocked — builds on the verified quotation record.
Pages for the black-letter rules that recur across thousands of decisions, clustered from citing sentences and grounded in quoted text. Alongside them: per-passage usage pages, treatment-over-time charts, and CSV/JSON export on every empirical-analysis page.
Design under way; builds on the full citation graph.
The citation graph, drawn: explorable maps of case-to-case influence, judge-to-doctrine relationships, and court families. Every visual ships with a readable text twin, so nothing lives only in a picture.
Awaits treatment signals at scale.
Cases that are still formally good law but whose treatment history — mounting criticism, courts declining to extend, narrowing distinctions — matches the pattern that has preceded past overrulings.
Committed direction, each waiting on a named dependency.
Awaits the corpus-currency program and email infrastructure.
Follow any case, statute section, judge, or issue. When new decisions cite it, treat it, or amend it, you hear about it — free, with an unsubscribe link that works.
Awaits broader state-code ingest; a topic pilot runs first.
Comparison matrices for the most-asked legal topics — statutes of limitations, non-competes, self-defense — across federal law and every state we host, every cell deep-linked to the statutory text it came from, with as-of dates.
Awaits the financial-disclosure ingest.
Cross-references federal judges’ financial-disclosure reports against the parties in the cases they decided, and surfaces the overlaps — with the underlying documents linked.
Builds on the statute-comparison machinery.
For the questions state courts answer differently — where does each state stand, which way is the trend moving, and which decisions did the moving. Built on the comparison machinery, sourced to the deciding cases.
Awaits attorney re-extraction across the corpus.
The attorneys and firms named in the record, connected to the cases they argued and the courts they argued them in.
Post-launch; ships with stable identifiers on our CDN.
The citation graph, the verified quotation record, and the treatment labels as documented, versioned datasets with stable identifiers — so research built on the record can be reproduced and cited.
Some things we've decided not to build: outcome prediction, docket analytics, and our own AI chatbot. The record is the product — we'd rather make it verifiable than make it talk.