Highlights

This release makes the Relativity integration faster in both directions, brings Case Insight onto our faster results pipeline, turns CaseBot 3.0 on by default, and gives clients a clearer, itemized view of what they're billed for.

• Faster Relativity integration: documents come in and results go back faster.
• Doc Viewer: see classifications and jump straight to the document in Relativity.
• Case Insight: now runs on the same high-throughput results pipeline as our other job types.
• CaseBot: CaseBot 3.0 on by default, cited-document lists in reports, and more reliable long-running skills.
• Billing: CaseBot moves to simple monthly billing, every charge is its own line, and partners get their own billing endpoint.

Relativity integration

Working with Relativity is now faster at both ends of a job, and reviewers can move between our results and Relativity in one click.

• Faster document pulls from Relativity. Documents are retrieved from Relativity faster, so jobs start processing sooner.
• Faster results back to Relativity. Results are written back to Relativity faster, so reviewers see them sooner.
• Classification in the Doc Viewer. The Doc Viewer now shows each document's classification alongside the document.
• View in Relativity. When a document lives in Relativity, a View in Relativity option opens it there directly.

Case Insight performance

Case Insight now runs on the same results pipeline our other job types moved to earlier this year, which has made those jobs run 3–4x faster than before.

• The pipeline scales with processing capacity: as we add capacity, Case Insight jobs get faster.

CaseBot

CaseBot 3.0 is now the default, reports show exactly which documents they cite, and long-running skills finish reliably even if the user steps away.

Cited documents in reports

• Skill PDFs and exported reports (PDF and CSV) now end with a Cited Documents list of every document referenced in the report.
• Fixed: citations in some Azure-hosted cases showed full storage URLs. They now show only the document filename.

More reliable long-running skills

• Skill results are saved even if the user closes the tab or loses their connection, and are there when they come back.
• Large skill runs get more time to finish, so fewer time out.
• Fixed: raw data no longer appears in the chat during large Request Analysis runs.
• Fixed: a response that was cut off no longer shows as complete when the chat is reloaded.

Polished skill reports

• Skill PDFs have a cleaner layout: fewer page breaks, no tables running off the page, and an updated logo.
• Every skill report now notes that document retrieval is ranked, not exhaustive.
• Questions can now be up to 10,000 characters (up from 2,000), so larger Request Analysis batches fit in one go.

Per-case AI settings

• Each case keeps its own saved AI configuration, so results stay consistent for the life of the case.
• If a case can no longer use CaseBot 3.0, the chat asks the user to reload rather than quietly switching to classic CaseBot.

Billing and usage

Billing is simpler and more transparent: CaseBot moves to monthly billing, every charge appears as its own clearly named line, and clients can pull their own billing records.

CaseBot is now billed monthly

• CaseBot is charged once at go-live, then monthly for hosting while it stays active. This replaces the 90-day extension cycle and its reminders.
• Purging CaseBot stops hosting charges from the following month on. The month in which it's purged is billed in full.

Every charge is its own line

• Case Insight, CaseBot deployment, CaseBot hosting and Case Elements each appear as separate, clearly named lines, so clients see exactly what they're paying for and when.
• Case Elements now appears as its own line rather than a flag on the job, with safeguards so a job is never charged twice for it.

Billing endpoint for clients

• Clients can pull their billing records from a dedicated billing endpoint using their API key.
• It shows the same charges as the Hub usage page, so both views agree.
• Results can be filtered by date range, job type, workspace, matter number and Relativity instance.
• Each record shows a readable job type name (for example, Case Insight) alongside the job type code used for filtering.

Hub usage page fixes

• Total results and page counts are now accurate. They were previously slightly high.
• Document unit figures now match exactly between the Hub and client billing data.