Legora AI Review: Collaborative Document Review for UK Legal Teams

JSH LAW / LEGAL AI PRODUCT RESEARCH

England and Wales perspective · Sources reviewed 23 October 2026 · Documentary assessment

From Leya to Legora: the product being assessed

The supplier announced the change from Leya to Legora on 19 February 2025. Older references to Leya therefore need to be read in their historical context, rather than treated as a separate current product comparison.

The current platform is positioned for law firms and in-house legal teams, with research, document interfaces and agentic workflows. “Agentic” describes software carrying out connected steps; it does not tell a buyer which steps require approval or how reliably the system handles exceptions.

The practical distinction from a general chat tool is the proposed organisation of shared work: document collections, repeatable review structures and interfaces used by a team. That is potentially valuable, but also means the firm must manage shared assumptions, access and publication of reusable instructions.

Why Tabular Review is the feature to examine closely

Legora’s Tabular Review documentation describes documents arranged as rows and prompts as columns. A cell can be expanded to inspect supporting material. The page also describes collaborative controls including review status and locked cells.

That creates a useful unit of supervision: an individual answer to an individual question about an individual document. A reviewer should be able to identify which cells are generated, checked, disputed or no longer current. A spreadsheet-like appearance alone does not establish those distinctions.

The table is a practical evaluation framework, not a report of product performance. Include empty and ambiguous answers in the test. A system that says “not established from these documents” may be more useful than one that fills every cell confidently.

Drafting, research and the document-management boundary

The supplier presents a Word add-in, research and structured workflows alongside Tabular Review. Confirm which capabilities are in the proposed licence and whether the firm’s exact document-management system is supported. Importing an isolated file is different from maintaining a live connection to a matter repository.

For drafting, ask for a demonstration with existing tracked changes, internal comments and an amendment that changes a defined term. For research, require an answer grounded in an identified source collection. For collaboration, ask a second reviewer to resume the work and establish what has been approved without an oral handover.

Exports deserve their own test. A review table that becomes an untraceable spreadsheet outside the platform may not meet the team’s needs. Request a sample export with source locations, qualifications and review status, and establish what a recipient can access without a licence.

England and Wales coverage needs a source-level answer

Legora’s international positioning is relevant to UK practices, but a multinational customer list does not prove comprehensive domestic coverage. Ask for the source list available in the proposed tenant and the update arrangements for legislation, judgments and procedure.

The question is particularly acute where one workflow combines the firm’s documents, legal research and open-web material. Those are different evidence classes. A negotiated precedent records what an organisation once accepted; it does not establish the current law. A general web article may explain an issue but cannot silently replace the controlling authority.

A useful UK trial should include a recent change, an older authority that remains relevant, an overseas decision that looks superficially similar and a document with an uncertain governing-law clause. Require the system to state its jurisdiction and identify uncertainty. English drafting style is not evidence of English legal coverage.

Three hypothetical uses with different risk profiles

A due-diligence team dividing a contract collection

Use fifty invented leases and side letters. One reviewer checks break rights, another assignment restrictions and another rent-review assumptions. The pilot should test whether a side letter changes the answer, whether the correct source remains attached and whether a later document update invalidates a previously reviewed cell.

The benefit being examined is shared review discipline. The unacceptable result is an apparently complete table that conceals an unprocessed side letter or preserves an old answer after its supporting text changes.

An in-house team applying an approved playbook

Use a fictional procurement policy, a standard agreement and a set of proposed amendments. Ask for a list of departures and suggested drafting, with the approving role for each material deviation. Include one clause requiring a senior exception decision.

This tests whether the workflow supports delegation without turning a sales team’s request into automatic legal approval. A suggested concession should remain visibly provisional until an authorised person accepts it.

A litigation team organising conflicting records

Use invented witness accounts and contemporaneous emails. Ask for a table distinguishing allegation, agreement, professional opinion and judicial finding, with exact quotations where necessary. Do not ask the system to decide which witness is truthful.

The potential value is a better map of the dispute. The limit is equally clear: categorising text is not authenticating it, establishing admissibility or making findings. For family proceedings, the trial must remain synthetic unless the legal and information-governance basis for live processing has been assessed.

Security claims, retention and the Portal problem

Legora’s security disclosures describe encryption, SSO, access controls, audit-related governance and configurable retention. They state that customer data is not used to train or fine-tune models. Some controls are described as enterprise-package options, so do not assume they are present in every commercial configuration.

The Portal announcement describes controlled collaboration with clients. That raises a distinct question: exactly which matter, file, output or workflow is being exposed to the recipient? Review a client-facing permission set from the recipient’s perspective, rather than relying on an administrator’s view.

Statements about model-provider retention should not be read as a promise that the platform stores no documents or history. A collaboration service necessarily needs a defined account of what persists. Obtain the customer agreement, processing terms, subprocessor list, support-access arrangements and exit schedule. Ask how deletion affects shared outputs and exported copies.

We have not inspected the underlying certification reports or tested whether any particular deployment satisfies UK GDPR. A certification is evidence to examine within supplier due diligence; it does not decide the lawfulness of uploading a protected family-court record.

Cost and implementation: when collaboration pays for itself

We did not verify a universal public UK price in the material reviewed. Obtain a written quotation identifying seats, modules, source subscriptions, integrations, implementation and minimum commitment. Confirm whether external collaboration and advanced governance controls cost extra.

The business case should measure more than faster first-pass extraction. Consider time spent reconciling reviewers’ answers, finding the source, correcting a mistaken assumption and producing the final client deliverable. If the new system merely moves that work to a different interface, the headline saving is misleading.

A small team with one-off documents may be better served by a narrower drafting tool and a disciplined manual register. A larger team with recurring document classes may gain more from shared templates and review controls. Neither conclusion requires a claim that one product is universally better.

How to run a fair Legora pilot

Start with a human-checked document inventory and a private answer ledger. Seed the sample with an illegible scan, a duplicate, a later amendment and an issue that the documents do not resolve. Define what counts as a critical failure before seeing the output.

Then test extraction, collaboration and export separately. Ask two users to disagree about a cell, record the resolution and revisit it after a source changes. Test an account that should not have access to the collection. Finally, export and archive the work, then establish whether another reviewer can reconstruct the answer.

Keep a record of prompts, source versions, configuration and human corrections. Repeating the task with a materially changed input is more informative than repeatedly asking for reassurance that the first answer was correct. These are proposed tests; no result is asserted here.

Alternatives and JSH Law’s assessment

Harvey is a relevant comparison for broader professional workflows. Spellbook is narrower where contract drafting inside familiar editors is the main requirement. Everlaw belongs in the discussion when the real problem is defensible litigation disclosure rather than ordinary document collaboration.

Our assessment: Legora’s table-based review and shared workflow proposition warrants serious evaluation by teams with repeated document-analysis work. It is not an automatic recommendation for an unrepresented court user. The decisive evidence would be reliable source tracing, visible reviewer responsibility and a manageable total cost on the buyer’s own task.

Is Legora a substitute for a lawyer’s judgment?

No. A shared table can make supervision easier, but deciding relevance, legal effect and the appropriate action remains a human responsibility.

Is a large document capacity enough to justify the purchase?

No. Processing success, retrieval completeness, permission controls and usable exports matter as much as the size of the advertised collection.

Sources and methodology

Sources reviewed on 23 October 2026: Legora’s platform, Tabular Review, security disclosures, Portal announcement and rebranding notice. The supplier’s detailed customer contract and full source entitlements were not independently established. Reassess after changes to connectors, review controls, sources or commercial terms.

The assessment is independent editorial analysis of documentation. It is not a hands-on benchmark, security certification or verified time-saving claim. See the research hub for the comparison directory and common evaluation method.

Where your actual need is organising evidence or checking a court-document draft, JSH Law provides defined-scope document-preparation support. Book a Consultation with the immediate issue and next deadline, rather than uploading a full confidential file to an enquiry form.