Foundry research · Great Library of SISO
Begin with supervised intake, approved-template assembly and evidence-preserving document review. Compare against the firm's existing software before replacing it. Neither exposure estimates nor a repository's features establish cash savings.
Research dated 2026-09-05. Business value remains unmeasured. Software descriptions are documented capabilities; each project must verify fit, rights and outcomes before adoption.
United States research pilot; state/practice-area requirements still need an authorized firm-specific review.
An occupational task inventory identifies legal-assistant document preparation, records organization and research. A preregistered 2024 study found errors in the legal AI research tools it tested despite retrieval grounding. This supports decomposing tasks and measuring supervised assistance; it does not supply a law-firm-wide automation rate or a current vendor-performance ranking.
Collect and normalize supplied fields; flag missing facts and possible matches; never accept a conflict or engagement automatically.
Populate approved templates from validated interview answers and return a draft plus field provenance.
Extract candidate facts with document/version/page references; reviewers accept/reject each finding.
Prepare a signing envelope only from approved immutable content; require explicit release authority.
Draft status updates from current records; route for review and send only when authorized.
Surface missing time descriptions, inconsistent identifiers or rejected line items in a read model; keep financial authority upstream.
Vendor currently advertises intake, case management, document automation, time/billing and client communication features.
No verified billing-basis quote or client invoice; do not count the subscription as cancellable.
Vendor currently advertises matter/workflow management, intake/conflict checking, document automation, billing and client portal features.
Firm-specific quote and actually avoidable spend unknown.
Current case-management product explicitly targets civil legal aid, public defenders and government law offices.
Do not treat a public-interest legal product as a like-for-like incumbent for a private commercial firm.
Guided interviews and template-based PDF/RTF/DOCX document assembly.
Document annotation, structured extraction and citation-linked review, as described in upstream README.
Self-hostable electronic-signature workflow, described by its upstream project.
Typed reusable connector pieces and workflow execution.
PostgreSQL-backed tables, forms, permissions and API advertised upstream.
No firm baseline, invoices, achieved automation rate, review/rework time, operating cost, displacement proof or eligible-population measurement. Null is not zero.
Within one authorized firm, compare matched routine work under the incumbent workflow and supervised candidate workflow. Start with synthetic/redacted fixtures; pre-register which documents/actions are in scope before using live data.
Observed task inventory, not an automation forecast or firm baseline.
Submitted 30 May 2024; the tested historical systems are not a current-market ranking. Supports verification/quality gates, not a savings percentage.
Capabilities observed on live vendor page; no firm adoption or performance measurement.
Capabilities observed on live vendor page; not a migration or savings proof.
Vendor explicitly distinguishes its buyer from for-profit law firms.
Deterministic assembly capability is not evidence that a generated legal document is correct.
Commit identity verified using GitHub API; README review is not executable module qualification.
Commit identity verified using GitHub API; README review is not executable module qualification.
Commit identity verified using GitHub API; README review is not executable module qualification.
Commit identity verified using GitHub API; README review is not executable module qualification.
Commit identity verified using GitHub API; README review is not executable module qualification.