Harvey vs Legora: research sources, document review, agents and price

Harvey, Legora and Gavel Exec in one table

Harvey and Legora describe themselves in nearly the same words. A platform evaluation turns up "the operating system for legal and professional services" on Harvey's company page and "The agentic operating system for legal work" as Legora's tagline. The table shows where their published details differ, and the fourth column is Gavel Exec, which you can try before either demo.

Vault or Tabular Review: reviewing a document set

Both products turn a pile of documents into a table of answers. Harvey publishes a ceiling, 100,000 files per vault, and Legora publishes none, saying only that Tabular Review handles "thousands of documents."

Harvey's Vault is a storage layer first. It holds files, email correspondence and past queries, pulls documents in from iManage, SharePoint or Google Drive, and can be opened to colleagues and outside parties. Test the review tables, which run on top of the vault and extract the same data points from thousands of documents into a single grid.

Legora's Tabular Review starts from the grid. You import from a document management system, an internal repository or a virtual data room, and every document becomes a row with your prompts as the columns. Any cell opens to show its reasoning and the source passage. Legora adds reviewer controls that Harvey's Vault page does not describe: Mark as Reviewed, Lock Cells and a Review Mode, plus reusable templates.

Agents: a team of agents or one Agent with Skills

Harvey sells agents in the plural and Legora sells one Agent. Harvey says it can "deploy a team of agents that split large tasks, work in parallel, and return review-ready outputs," while Legora describes a single Agent that plans the work, carries it out, checks its own result and delivers it.

Harvey lets a lawyer preview the plan, change the scope and approve the work before it starts. Agents can run on a schedule and return reports or presentations. Agent Builder creates custom agents that draw on your knowledge, playbooks and preferences, and Harvey states that every step is logged.

Legora's Agent proposes a plan and picks its own tools. Skills are the teaching mechanism: they "encode your standard, processes, and preferences so the Agent works the way your team does." Legora also describes the Agent as proactive, saying it "monitors your email, documents, and projects," and it produces editable Word, Excel and PowerPoint files.

Harvey is ahead on published controls for scheduled and parallel work. Legora is ahead if you want a single agent that watches matters and email and speaks up first.

Research: named publishers or unnamed partnerships

Harvey names its research partners. Legora describes its coverage and names no publisher. That difference matters most to a firm that plans to rely on either product for case law.

Research evaluators get partner names from Harvey's Knowledge page, which promises "a powerful data pipeline into 500+ legal sources across the globe": Wolters Kluwer, Ask LexisNexis, Lefebvre, SCC Online, Otto Schmidt, Tirant lo Blanch, FromCounsel and Gyldendal Rettsdata. It also connects to EDGAR, the Securities and Exchange Commission's filing database, and to EUR-Lex for European Union law. Put one line in the evaluation budget: Harvey's help center says "Ask LexisNexis is licensed separately from Harvey's Basic Service," which makes LexisNexis primary law for the United States an add-on.

Legora's Legal Research page describes "comprehensive legal data coverage from publisher partnerships and official sources," with attorney editors verifying accuracy, a goal of hundreds of jurisdictions and a United States corpus it puts above 60 million documents. The page names no publisher, and its integrations page lists iManage and Intapp tools, with no research database among them.

Ask Legora which publishers sit behind an answer and whether a citator is included. Ask Harvey what Ask LexisNexis adds to the quote.

Word and Outlook add-ins

Both vendors ship a Word add-in and an Outlook add-in, and both run playbooks in Word. Harvey documents its Word behavior in more detail, including a platform caveat.

Suggestions a lawyer accepts in Harvey for Word "are applied as redlined tracked changes in the document," and where a clause departs from your positions, Playbook review flags it. Evaluators on a Mac should know the help center points them to Word Online and recommends "a Windows computer with Harvey for Word" for best performance. In Outlook, Harvey summarizes threads, drafts replies and routes emails to Vault, and it has native iOS and Android apps.

Legora's Word add-in brings the Agent into the document so you can "draft, redline, and review without leaving your document." Playbooks are defined once in Legora, with starting positions, non-negotiables and fallbacks, then run in Word. Its page does not use the words "tracked changes" or name Windows or Mac, so confirm both in the demo. In Outlook, Legora summarizes long threads, writes replies and files emails to a matter, and it lists an Editor and a mobile app as well.

Which AI models, and who controls the keys

Harvey is more specific about models and regions, and Legora offers one control Harvey's public security page does not mention.

Harvey's help center calls the product "multi-model by design" and says it uses "models from Anthropic, OpenAI and Google, with Mistral also available in the EU." Its security page tells a reviewer that Harvey hosts on Microsoft Azure, "requires Zero Data Retention (ZDR) by model providers" and bars those providers from training on customer data. Customers can keep data in the United States, Australia, or the European Union or Switzerland, and Harvey is audited every year for SOC 2 Type II and ISO 27001 and shows badges for ISO 27701 and ISO 42001.

Legora's security page says it is certified to ISO 27001 and ISO 42001 and has a SOC 2 Type II report. It encrypts data in transit and at rest, offers customer-managed keys ("our BYOK option"), runs "semi-annual penetration tests covering the full platform scope" and commits that customer data is never used to train or fine-tune models. The page names no model provider, hosting company or region, and it does not use the phrase zero data retention. A customer story published by Anthropic says Legora uses Claude throughout its platform.

Harvey answers the model and region questions on its website; Legora's security page leaves them for you to ask. Our security questions to ask legal AI vendors give you the list to send both.

Price, minimums and how you buy

Neither company posts a price list, and neither offers a trial on its website. The only figure either vendor publishes sits in Harvey's help center: "On average, Harvey costs several hundred dollars per user per month." For Legora, the number reported by Gavel customers is about $3,000 per user per year with a $30,000 minimum, or $250 per user per month, and Legora's own site says it has no seat minimum.

Expect a demo first and a quote second, which Harvey says is how most teams begin, since in its words "pricing depends on the number of seats, which products you deploy, and contract terms." It publishes no seat minimum and no contract length.

On minimums, Legora's own site and the reports from Gavel customers differ. Legora's page for small and mid-sized firms answers its own seat-minimum question with a flat "No" and says a firm can buy "from 1 license to 5+." Gavel customers report a $30,000 floor.

For five lawyers for one year, Legora's reported rate comes to $15,000, or $30,000 if the reported minimum applies. Harvey's average gives no exact figure. Every $100 per user per month is $6,000 a year for five seats, so $300 per user per month would be $18,000 and $500 would be $30,000, against $250 per user per month at Legora's reported rate. Gavel Exec at $1,740 per user per year is $8,700 for the same five lawyers.

Firm size, customers and funding

Both report large-firm customers, and both now publish pages for smaller firms. Harvey reports the larger footprint.

Harvey's help center counts "3,500+ customers in 85+ countries" and its September 9, 2026 funding announcement says 80% of Am Law 100 firms use it. That announcement reported a $550 million round at a $15.5 billion valuation. Harvey's page for boutiques and mid-sized firms of about 15 to 150 lawyers lists Keker, Van Nest & Peters, Gray Reed and Moses Singer among its users, and it says setup and training are carried out with Harvey's support.

Legora's Series E, announced October 21, 2026, raised $550 million and valued the company at $8.8 billion. It reports more than 140,000 legal professionals at over 2,400 customer organizations in more than 80 markets, counts over half of the AmLaw 50 as customers and calls the United States its largest market by revenue. Its small-firm page says firms are up and running within days.

Which should you choose?

• Choose Harvey if you need research from named publishers including LexisNexis, matters that can reach 100,000 files, data kept in a named region, and a vendor that tells you which AI models it runs. Budget for several hundred dollars per user per month plus research add-ons.
• Choose Legora if you want one Agent across matters, Word and Outlook, reviewer controls in Tabular Review, customer-managed encryption keys and a price Gavel customers report at about $3,000 per user per year with a $30,000 minimum (Legora's site says it has no seat minimum), and you are willing to confirm the minimum and the research sources during the sale.
• Consider Gavel Exec if most of your work is negotiating contracts in Word and you would rather start today on 25 free queries than sit through two demos. The price after that is $1,740 per user per year, or $160 per user per month, with no long-term contract.

A third option for contract work: Gavel Exec

A firm that cannot or does not want to commit to a platform contract sold by demo can open Gavel Exec today on 25 free queries per user, with no credit card and no sales call. It is AI contract review and drafting software built for transactional attorneys, in-house legal teams, and solo, small and mid-sized firms whose week goes to marking up the other side's paper.

• Price and trial: once the free queries are used, paid plans run month to month at $160 per user per month, or $1,740 per user per year, and include 1,000 queries per user per month.
• The redline: Gavel Exec redlines the entire agreement inside Word. The edits arrive as tracked changes ranked by priority, each with its reasoning, along with margin comments and a list of missing terms.
• Playbooks: Gavel Exec will generate rules from your checklist, spreadsheet playbook or signed agreements, and it includes built-in playbooks developed with attorneys in corporate and real estate law. Share either kind read-only or with edit access.
• Benchmarking: the web app labels every provision on-market, off-market or missing, filterable by company size, industry and jurisdiction.
• Security: Gavel Exec is SOC 2 Type II certified, and Gavel holds Zero Data Retention agreements with the AI model providers it uses, OpenAI included. Every customer, your firm included, gets its own subdomain, isolated database and separate file storage. See the Gavel security page.

At MIA Contract Lawyers in Australia, Martin Algie reduced first-cut review from about 15 hours per contract to about 5 hours. Omar Nakadi, Legal Counsel at a regulated financial services firm, estimates that each contract takes 40 to 60% less review time. More than 2,000 legal organizations in 28 countries use Gavel, which was founded by Dorna Moini, its CEO and Founder and a former litigation associate at Sidley Austin.

Because Gavel Exec is built for transactional work, it has no case-law research database or native iManage or NetDocuments integration today; its source checks draw on your precedent, matter files, SEC EDGAR filings and government sources, and litigation and e-discovery are covered by Relativity, Gavel's parent company. The head-to-head detail is in Harvey vs Gavel Exec or Legora vs Gavel Exec.

How to test Harvey and Legora side by side

1. Pick three real tasks. Choose one research memo, one data room of 50 to 200 documents and one contract you have already negotiated.
2. Book both demos in the same week. Neither site offers a trial, so ask each vendor for a pilot workspace.
3. Run the data room. Build the same ten columns in a Vault review table and in Tabular Review. Open twenty cells and check each citation against its source.
4. Check the research. Ask the same question of law in both. Note which publisher each cited case came from, and ask Harvey whether the answer needed Ask LexisNexis.
5. Redline the contract in Word. Run your playbook in each add-in on the machines your lawyers use, Mac included, and compare the changes with the markup you sent.
6. Get five numbers in writing. Per-seat price, minimum commitment, contract length, renewal notice period and the cost of research add-ons.
7. Run the contract through Gavel Exec the same day. The 25 free queries need no sales call, and the result tells you whether contract work needs a platform purchase at all.

More comparisons: Harvey alternatives, Legora alternatives, Spellbook vs Harvey, Spellbook vs Legora, what AI contract review software costs and every comparison Gavel publishes.