Return Editor

Return Editor works on folders of contracts stored on the user’s own computer as plain Markdown files, with AI processing performed locally unless cloud processing is selected.

Key features and functions include:

Document Import
Imports PDF, Word, OpenDocument, RTF, and HTML files and converts them to Markdown while retaining the original file. Clause numbering, tables, footnotes, and images are preserved during conversion.

Returns
Provides reusable AI analysis templates that can be applied to any document, with parameters such as jurisdiction, party perspective, or house rules. Built-in analyses include risk assessment, summaries, key date extraction, and plain-language notes.

Traffic Lights
Marks clauses in place as green, amber, or red, with an explanation of why each clause was flagged and which party it affects.

Playbook Check
Checks a contract against a firm’s own playbook position by position, with the quoted text supporting each result.

Chat and Agent
Answers plain-language questions about the open document or the whole workspace, with citations linked to the document’s own section numbers. The agent reads the document section by section and can draw on passages from other files in the folder.

Semantic Search
Finds passages by meaning across all documents in the workspace, using a search index built and stored on the user’s computer.

Stacked Edits
Shows every AI change as an inline diff with the original text struck through. The user accepts or rejects each change, with undo available.

AI Suggestions
Drafts three alternative wordings for a flagged clause, ordered with the strongest option first, which can be applied in one step.

Placeholders
Converts documents into reusable templates with fill-in fields.

Document Constraints
Applies user-defined rules for tone, format, and style to every AI edit.

Formulas and Charts
Supports formulas and charts within documents, for example to model fee caps or damages scenarios.

Export
Exports documents to PDF and HTML.

Local and Cloud AI
Includes a local AI engine that runs on the user’s computer and recommends a model suited to the available hardware. In local mode, documents, prompts, and responses remain on the computer, and the application works offline without an account. Cloud AI can be selected as an option, and cloud requests are not used for model training.