Litera adds Midpage research to Lito in Microsoft 365

Litera has agreed to integrate Midpage into its AI legal agent, Lito, bringing US case law and statutes into the tool. The announcement came alongside internal benchmark results on how large language models perform on complex document comparison tasks.

The tie-up embeds Midpage's legal research in Lito within Microsoft 365, where many lawyers draft and manage correspondence. Users will be able to query statutes and case law alongside their own documents through Lito's chat interface.

The companies cast the integration as part of a broader shift toward tools that sit directly inside everyday legal workflows. Litera also argues that general-purpose large language models struggle to produce the structured outputs lawyers exchange during negotiations.

"Every legal AI tool has access to the same foundation models," said Adam Ryan, Litera's Chief Product Officer. "The difference is what surrounds them. Lito combines the best large language models with our rules-based engines, cutting edge firm intelligence data, and now deep legal research - all integrated where lawyers already work."

Research in workflow

Midpage is a legal research platform used by more than 200 law firms, according to the companies. The integration gives Lito access to US statutes and case law as selectable sources when a user asks a question or analyses a document.

Litera pointed to use cases such as checking a contract against a statute, uploading a document with relevant authority for analysis, or generating a case summary for client communications. The feature sits inside Word and Outlook for users working in Microsoft 365.

The legal research capability is available to Lito users on Litera One cloud packages. Customers can extend access through a Midpage subscription.

The agreement also expands Litera's integration programme, which includes more than 60 integrations with products such as NetDocuments, iManage, Courtroom Insight, and UniCourt.

"Navigating case law has historically been so complex that it was really only done for complex litigation," said Otto von Zastrow, Midpage CEO. "AI agents give every attorney the power of a big legal research team. The agent reads hundreds of cases and finds on-point precedents with quotes and hyperlinks. We're glad to bring this to tools like Lito that already have access to your documents and important context."

Benchmark results

Alongside the integration, Litera shared details from internal quality engineering work comparing Litera Compare with several general-purpose large language models on legal redlining tasks. Models included Gemini 3, Claude Opus 4.5, and ChatGPT 5.2.

Redlining is a core part of contract negotiation and review. Lawyers rely on tracked changes and standardised comparison files both as a record of edits and as a format for exchanging revisions with counterparties.

The tests used long-form legal documents containing elements that often complicate comparisons, including tables, images, embedded objects, headers and footers, footnotes, and other structural components.

Litera said general-purpose models did not produce usable redlines for non-text elements such as tables and images. It also said the models could describe changes but could not generate a redline or tracked-changes file suitable for exchanging with other parties.

Litera also reported accuracy problems as documents grew larger. Even on shorter documents, it said general-purpose models reached about 90% accuracy-too low for legal work, where a single missed change can alter meaning or shift risk allocation.

In a test with a 200-page document, Litera said one model's text accuracy fell to about 40%, while others dropped to roughly 70%. Performance also varied by model on longer, more complex files, with some completing comparisons faster but producing less reliable or less complete output.

Compare and Lito

Litera Compare is the product used for document comparison and redlining. Compare underpins redlining in Lito, allowing the agent to return comparison outputs in a format lawyers can use in negotiations.

Litera describes its approach as combining large language models with deterministic rules-based engines and firm intelligence. That architecture, it said, matters when the work requires repeatable formatting and an auditable trail of changes.

Adding Midpage brings another source of context into the same interface, combining research with drafting and review. For some firms, the package reflects a shift away from standalone research tools and separate drafting assistants toward consolidated systems inside Microsoft 365.

Litera and Midpage are presenting the integration and benchmark findings at Legalweek in New York. Litera also plans further discussions on how legal AI is being assessed against measurable performance in document work.