Legora

Legora is buyer-operated software for legal teams rather than an outsourced legal service. Its public Workflows page describes a configurable system that combines drafting, review, extraction, search, and custom tools, drawing on reference documents, precedents, and style guides. It gives supplier-agreement analysis, compliance review, employment-agreement review, and lease red-flag reporting as examples. [S1]

The best fit is a legal team that already has approved positions, precedent, a qualified owner for exceptions, and a repeatable document workflow worth testing. The Word Add-in page says Legora can draft, redline, and review within Microsoft Word and lets teams define playbooks with rules, starting positions, non-negotiables, and fallbacks. [S4]

The important trade-off is configuration and governance: a useful workflow needs clear source material, tested permissions, defined escalation, and legal review of material outputs. Public pages do not establish the buyer's exact integrations, permissions, legal-content coverage, data path, pricing, retention, support terms, or exit rights. Confirm those in the proposed tenant and agreement.

Best for: in-house legal teams and firms with recurring document work, settled playbooks, accountable reviewers, and a controlled rollout plan.

Not for: a team seeking ungoverned automated legal advice, without an owner for playbooks and exceptions, or unable to verify data, privilege, access, and export terms for the documents it plans to use.

Review repeatable agreements against a playbook

• Trigger: A team receives a routine NDA, supplier agreement, employment agreement, or comparable document that fits an approved review policy.
• Inputs: The document, buyer-approved starting positions and fallbacks, relevant precedent, matter facts, jurisdiction, and a defined exception route.
• Output: A structured analysis, proposed redlines, a red-flag report, or identified questions for the responsible team.
• Human checkpoint: A qualified legal owner checks material provisions, source relevance, exceptions, and the final negotiating position before any external commitment.
• Success measures: independently reviewed material-issue coverage, legal-owner overrides, false positives, missed issues, exception age, turnaround, and rework.

Legora lists supplier-agreement analysis, employment-agreement review, and lease red-flag reporting among its workflow examples. The public examples establish possible workflow shapes, not coverage for every document, jurisdiction, or buyer policy. [S1]

Draft or redline in Microsoft Word

• Trigger: A lawyer or authorised team member needs a new clause, a revision, a redline, or a review within a Word document.
• Inputs: The Word document, a playbook with approved rules and fallbacks, permitted reference material, and the factual and commercial context.
• Output: A draft, redline, review notes, or a proposed language change in the document.
• Human checkpoint: The lawyer validates legal and factual fit, approves any departure from the playbook, and controls outbound communications and signature.
• Success measures: accuracy against approved positions, citation or source accuracy where applicable, edit acceptance rate, material-rework rate, and time to approved draft.

Legora says the Word Add-in supports drafting, redlining, and review without leaving the document; it also says Playbooks can codify rules, starting positions, non-negotiables, and fallbacks for documents such as NDAs and service agreements. [S4]

Apply controlled edits across a document set

• Trigger: A legal team needs a consistent, repeatable change across an approved set of Word documents.
• Inputs: The approved instruction, source or master document where relevant, target documents, version and jurisdiction boundaries, and sampling rules.
• Output: Updated Word documents for review, or an exception list where the requested change does not safely apply.
• Human checkpoint: The workflow owner approves the instruction and scope; a legal reviewer samples results and handles non-standard, local, or high-risk documents.
• Success measures: correct-edit rate, missed-document rate, formatting defects, exception detection, sample-review disagreement, and correction time.

Legora's February 10, 2026 Word Edits announcement describes using the Assistant or Workflows to edit multiple .docx files, fill templates, generate ancillary documents, standardise language, and maintain local or market-specific versions. Treat that as product capability evidence; validate the buyer's document formats, permissions, controls, and actual output quality. [S2]

A conservative operating model is: an authorised buyer user selects an allowed matter and inputs → Legora workflow reads only the permitted documents, precedents, and instructions → it produces a draft, review, extraction, or structured analysis → the responsible lawyer checks sources, factual fit, legal reasoning, and exceptions → authorised legal and business owners approve positions → an authorised person sends, files, or signs → the buyer retains the required work record.

Legora says workflows can be built from its tools, reference documents, precedents, and style guides, and that teams can use flexible permissions and role-based controls. Those are vendor-described product capabilities, not evidence that a particular tenant has a suitable permission or audit configuration. Test roles, content access, workflow editing, output sharing, exports, logs, revocation, and emergency shutdown in the buyer's environment. [S1]

The U.S. terms define the service as a browser-, mobile-, desktop-, API-, plug-in-, and add-in-accessible legal-AI platform when those elements are applicable under the order form. They say Legora processes personal data in subscriber content on the subscriber's behalf under its DPA, and state that subprocessors other than cloud-storage providers must not retain or log subscriber content or confidential information for human review. The order form, DPA, subprocessor list, region, and product configuration remain decisive. [S3]

Verified facts from public primary sources

• Legora's Workflows page says its workflows combine drafting, review, extraction, search, and custom tools; it names reference documents, precedents, and style guides as workflow inputs. The page dates Workflows to June 24, 2025. [S1]
• The Workflows page gives internal compliance review, supplier-agreement analysis, employment-agreement review, lease red-flag reporting, and other legal tasks as examples. [S1]
• Legora's February 10, 2026 announcement says Word Edits can make changes across multiple Word documents through the Assistant or Workflows. [S2]
• Legora's May 2026 U.S. terms define the service as a legal-AI platform delivered through interfaces and add-ins described in an order form and set out statements on subscriber-content processing. [S3]
• The Word Add-in page says teams can create Playbooks with rules, starting positions, non-negotiables, and fallbacks for review of documents including NDAs and service agreements. [S4]

Vendor claims that need buyer validation

• Legora says workflows are agentic and can execute complex legal tasks step by step. The buyer should test whether a selected workflow reasons over the correct source material, reaches usable conclusions, flags uncertainty, and routes exceptions to people. [S1]
• Legora says its Word Add-in can support accurate and consistent drafting, redlining, and review. Measure performance on the buyer's own precedents, matter mix, and difficult legal and factual cases rather than assuming that outcome. [S4]
• Legora says Word Edits is suitable for repeatable, high-volume document work. Confirm document formats, result traceability, reviewer controls, and the treatment of exceptions before using it for consequential changes. [S2]

B2Bagents assessment

Legora is most credible as a legal-workflow layer for repeatable work that stays under lawyer and business-owner control. The visible capabilities can reduce manual drafting and review effort, but they do not replace the need to establish which sources are authoritative, when a workflow must stop, who approves deviations, and how sensitive document content is governed. A pilot should show that the buyer's approved positions survive the workflow, hard cases surface clearly, and human review remains real rather than ceremonial. [S1][S2][S4]

Material unknowns

• Public pricing, packaging, user or consumption unit, legal-content licensing, implementation fees, support commitments, renewal, cancellation, data-export, and transition terms.
• Tenant-specific integrations; what each connector can read or write; role configuration; workflow-change approval; API scope; service accounts; log and audit-record detail; and emergency disablement.
• Applicable DPA, model and subprocessor list, data and support locations, encryption, retention, deletion, backup, legal holds, incident notification, assurance-report scope, access controls, confidentiality, privilege, and data-residency terms.
• Output quality, source coverage, jurisdiction coverage, language behaviour, hallucination or missed-issue rate, error explanations, correction workflow, legal-review controls, and treatment of novel, regulated, or conflicting matters.
• The buyer's ability to export documents, prompts, workflow definitions, source links, decisions, logs, and approved playbooks in a usable format at termination.

Start with one frequent, low-to-moderate-risk document type supported by an approved playbook and a named legal owner. Define the allowed source documents and the workflow's objective, inputs, access roles, output format, exception classes, legal and business approvals, outbound authority, record owner, correction process, and stop mechanism before feeding in live matters.

Use a representative comparison set containing both standard and difficult documents: changed or missing schedules; non-standard liability, indemnity, IP, privacy, security, data-use, cross-border, and governing-law terms; conflicting precedent; regulated facts; ambiguous instructions; role changes; wrong source documents; and revoked access. Run the workflow in parallel with existing legal review until a qualified reviewer can compare results without relying on the tool to judge itself.

Track legal-owner agreement, material-issue coverage, source or citation accuracy, false positives, missed issues, rework, exception volume and age, turnaround, approval completeness, inappropriate access or sharing events, correction time, and export completeness. Set thresholds before expansion. Stop or narrow the pilot after a material missed issue, an unsupported legal position, unclear source basis, unauthorised access or action, an untraceable data path, a failed correction, or inability to export the buyer's records.

• Harvey: compare a broad legal-AI environment with Legora on source coverage, document-workflow configuration, integrations, and the buyer's preferred operating model.
• Spellbook: compare Word-centred contract review and playbook control, especially on the document workflow, legal-team administration, and evidence needed for output quality.
• Luminance: compare contract-lifecycle and negotiation workflows with a legal-workspace approach, focusing on lifecycle integration, permission design, human review, and document-record ownership.

1. Which interfaces, modules, legal-content sources, integrations, and APIs are included in our edition, and what can each read, write, share, retain, or export?
2. How are playbooks, workflow definitions, source documents, prompts, outputs, citations, user actions, and decisions versioned, approved, logged, searched, corrected, and recovered?
3. Which data and access controls apply to our tenant: models, subprocessors, storage and support locations, encryption, SSO, RBAC, audit logs, retention, deletion, backups, legal holds, residency, and incident commitments?
4. How does a lawyer review or override a workflow; what blocks or routes high-risk work; and can we enforce mandatory approvals before sending, filing, or signing?
5. What evidence can Legora provide for our jurisdictions, document types, languages, legal-content coverage, Word and repository integrations, implementation support, and response commitments?
6. What are the commercial and exit terms for licences or consumption, onboarding, support, legal content, workflow limits, renewal, cancellation, records export, and transition?