Harvey Release Notes

Describe what you need in plain language and Harvey finds relevant documents across iManage, SharePoint, and Google Drive.

A lawyer needs to find how their team handled a similar issue in a past matter. The answer may be in an agreement, brief, or email, but they may not know the filename, folder, or even the right search terms. Finding it can mean asking colleagues where to look and opening files one by one.

With agentic search across iManage, SharePoint, and Google Drive, lawyers can describe what they need in plain language and search connected documents without selecting files first. Results link to their sources, so lawyers can review prior work in context.

Find Relevant Documents Without Knowing Where They’re Stored

With Harvey's new agentic search capability, you can simply describe what you need in plain language and Harvey plans the search parameters to find relevant documents without requiring you to select each file first.

A lawyer preparing an acquisition agreement could ask Harvey to find agreements with an explicit anti-sandbagging clause. Harvey searches connected Document Management Systems (DMS) and returns relevant results with citations linked to source files. The lawyer can review the language in the context of each agreement and decide what applies to the current project.

If more results are available, Harvey pauses for the lawyer to review what it has found when appropriate. Based on that feedback, it narrows the search or continues reviewing results, then brings relevant documents into the thread for the lawyer to work with.

Agentic search is available in Harvey Assistant and the Harvey for Word Add-in, so lawyers can find relevant documents while researching or drafting.

Narrow the Search to the Work at Hand

As part of our work to make your internal knowledge truly searchable, we recently released functionality that makes searching more precise. To narrow the results, lawyers can:

• Search within a specific site, folder, or drive
• Filter by document author, with help choosing the right person when names match
• Put a phrase in quotation marks to find an exact match, such as “anti-sandbagging”

Keep Existing Access Controls Intact

Legal teams often keep sensitive documents in folders or workspaces limited to specific people or matters. Those access controls continue to apply when someone searches through Harvey.

Harvey uses each person’s permissions in the connected system, so documents they cannot access there do not appear in their search results. Lawyers can search across connected content while the existing boundaries on who can see each document remain in place.

Put Prior Work to Use

Legal teams build up valuable knowledge over decades, but much of it stays buried in documents that are difficult to find later. Agentic search helps bring that work back into view.

A litigation team can find prior briefs and expert declarations while developing a new argument. A lawyer joining an existing matter can locate the documents and correspondence they need to get up to speed. When prior work is easier to find, teams can draw on it more readily in the matters ahead.

Agentic search across iManage, SharePoint and Google Drive is available now in Harvey. Talk to your Harvey representative about turning it on for your workspace, or request a demo to see it against your own content.

Today, we're introducing Custom Agents for Word

the same custom agents your team has built in Agent Builder, are now available directly in the Harvey for Word Add-in. Tasks such as NDA review, redlining, document summarization, and post-review memos use your review standards, preferred positions, review approach, and the output format directly in your open Word document.

Run Custom Agents Where you Draft

In Word, you can browse all your team’s custom agents in the Agents tab and pick the right one, or @-mention the agent you want directly in your query. The agent runs against the Word document already open, using its content as context. You can pull in any other document you need from Vault, and if the agent needs more information mid-run, it asks you directly in the thread instead of requiring every detail up front.

For example, for an initial master services agreement (MSA) review, you can open a counterparty's draft MSA in Word and run your team's own first-pass MSA review agent, the custom agent your team built to check the draft against its preferred positions. It flags deviations in liability caps, indemnification, termination, and data-protection terms, then returns the review in your team's format and house style. You get the same standardized first-pass review your team would perform by hand, now running on the open document in Word and reflecting how your team reviews MSAs.

Apply Your Judgment Before it’s Final

The agent returns a list of suggestions, each one naming the specific change it proposes — such as updating a renewal structure or conforming a fee to a negotiated term — and showing a preview before you apply it. You can use your judgment to apply a suggestion, apply it with a comment, or discard it. You can also apply all suggestions at once. When you apply a suggestion, Harvey inserts the edit as a tracked change, so you can accept or reject it in Word. If you need another pass, you can reopen the same thread later, so a review spanning multiple sessions doesn't start over.

Agents That Stay in Sync Across Harvey

There's no separate Word version of an agent to keep current. Any updates made to your agent’s instructions in the web app carry over the next time you run the agent in Word, applying the same review judgment and house positions in Word that your organization relies on elsewhere. Recently used agents also appear alongside your other Word workflows, including Playbooks, so you can find the right one again when you need it.

One Home for Every Word Tool

The Harvey for Word Add-in now organizes every Word tool into three tabs: Threads, Agents, and Library. Threads holds your chat sessions with Harvey, each one preserving its query history and document context. The Agents tab is where you can browse and launch custom agents. And the Library tab holds Harvey's built-in Word tools — Playbooks, Redact, Translate, Proofread, Convert to Template, and Fill Template — so your threads, the agents you built, and the tools you reach for all sit in one place.

Getting Started

Custom Agents for Word is now available*. To use Custom Agents in Word, you need the Harvey for Word Add-in, which you can install through Microsoft AppSource or through your IT team. Visit the Harvey Help Center for detailed setup instructions.

*This release supports custom agents created via the Agent Builder UI. Custom workflows will not be supported in the Harvey for Word Add-in yet.

Harvey helps lawyers find the cases that matter, understand how they relate, and apply that authority to their own work.

It's 11pm. You have a hundred and forty tabs open. You've read enough district court and appellate opinions that you could recite the doctrine by heart. You just need one court that agrees with your position. Most of what you're reviewing will not end up in the brief. You read more decisions anyway, because the only way to know which cases matter is to go through the ones that don't.

This is not busywork. Somebody has to explain the subtle distinctions that your client trusts you to catch — like the Northern District of Illinois applying a
slightly
different interpretation of the relevant issue than the Eastern District of California. Reading, understanding, interpreting, and mastering the relationships between cases is at the heart of professional legal research. But before you can develop mastery, you need to simply find the cases that matter.

Agents That Help you Build Mastery

This is where Harvey steps in. Harvey extracts the relationships across its entire case law corpus — which cases cite which, how they’re treated, where a nuanced line of reasoning holds and where it breaks – and makes them available to the agent as it works. So when you submit a query, the Harvey agent can pursue those connections itself, moving through the relationships between opinions before responding to your issue.

All of this work is in service of the lawyer’s own mastery over the case law. The Harvey agent iterates behind-the-scenes: retrieving authority, following relevant citations and distinctions, checking its work against the prompt, and returning grounded answers with inline citations. The agent reveals the shape of the authority on the issue, providing lawyers with a map for deeper understanding and professional expertise. The lawyer can focus on the harder part, like challenging the scope of holdings, finding key distinctions, or working out whether a case is actually fatal. This is deliberate cognitive work: mastery, not retrieval.

US Case Law in Harvey

We made Harvey’s US Case Law knowledge source available to customers in June, so research, analysis, and drafting could happen in one platform. Harvey’s US Case Law comprises millions of opinions across the entire federal judiciary and all 50 states – with coverage for over 2,100 federal and state courts, including more than 8 million published opinions along with scanned reporter copies. It includes comprehensive reported case law for all federal and state jurisdictions alongside unpublished decisions.

Harvey has also spent months indexing, extracting, and substantively enriching each opinion. The Harvey agent can now search by publication status, date filed, court, citing references, and cited by references. The agent can review Bluebook citations and isolate specific page breaks from reporters. All of this work results in more accurate, insightful, and valuable research outputs for US Case Law queries.

Harvey will continue adding to and enriching its US Case Law corpus, with greater coverage for unpublished decisions and state trial court orders in the coming weeks. US Case Law will expand even further as Harvey invests in gathering and digitizing more case law.

Mastery in Practice

The real measure of value for Harvey’s investment in US Case Law is whether lawyers continue to use and trust it. And in the span of a few weeks, US Case Law has quickly become one of our most popular knowledge sources on the platform. Users have more than doubled every month since our release in June. With increasing week-over-week growth and new records set each day, the Harvey agent performs cutting edge research in support of lawyers’ needs — from high-stakes filings to research emails.

Most importantly, lawyers

do

things with US Case Law. They attach documents to US Case Law queries at twice the rate of other sources, run more deep research tasks, query it more frequently in the Harvey for Word add-in, and click on citations more than twice as often as other sources. Harvey’s US Case Law allows users to accomplish more substantive work, reducing their time to mastery.

Lawyers can query about circuit splits and their impact on the fact pattern in their own filings. They can hand Harvey opposing counsel’s motion in limine and ask what authorities they rely on and the strongest cases in response. They can query US Case Law right inside the brief they’re drafting, through the Harvey for Word add-in, without leaving the document. For every answer, citations come back rendered inline, so the partner reviewing the argument can immediately read the relevant passage next to the claim.

These aren’t just hypothetical use cases. Lawyers are already performing all of these actions and more with US Case Law within Harvey. In the coming months, we expect US Case Law to become even more deeply embedded in lawyers’ daily routines, always decreasing their time to mastery.

What We’re Building Next

Lawyers are bringing substantive legal questions to Harvey’s US Case Law, attaching documents and context at a significantly higher rate than other knowledge sources. That’s why we’re investing even more here in the next quarter, with dedicated teams across Product, Engineering, AI Research, Data, Legal Engineering, and more.

Three pieces are underway. First, deeper classification to understand the structure of opinions. Lawyers ask complex questions, and the Harvey agent needs to know how key features relate to one another: which court, which judge, what procedural posture, scope of the disposition, subsequent history, and more. All of these elements will feed into the Harvey agent’s retrieval and analysis. So the agent knows that there’s a supporting proposition for your argument but the relevant court ultimately ruled the other way, and the agent is intelligent enough to flag these crucial nuances.

The second is precedential treatment: a warning on a case when the court’s opinion was ultimately overturned on appeal. But not a static warning that was pre-computed before your issue was even possible, meaning treatment removed from your specific facts and issues. Hard cases exist because of interpretive ambiguity, and precedential status is rarely binary for the issues that matter. So the Harvey agent will check at runtime whether the specific claim that you want to make remains viable, even amidst overturned, abrogated, questioned, or ambiguous language.

And the third is the brief itself: hand Harvey a finished draft to check every citation against its opinion. Features like “does the case exist” or “is the quote accurate” are just the beginning. Harvey will also explain whether the introductory signal is appropriate (see and but see are worlds apart), how the parenthetical’s description lines up with the case’s holding, and whether other cases tend to be cited alongside the one you chose. The Harvey agent will perform all of these actions and more — always in the interest of allowing you, the reviewing lawyer, to master the cases at hand.

We’ll be announcing more details at the Chicago Harvey FORUM in November.

Every action in a Review Table can now be delegated to the Harvey Agent in plain language, so the table can be reshaped, corrected, and routed as fast as the matter changes.

A litigation team finishes building a review table across hundreds of documents in a discovery production. The "privilege designation" column needs to be reclassified to Attorney-Client, Work Product, or Not Privileged, instead of a free-text answer. The Relativity export sitting in team members’ inboxes has custodian metadata that should map into the table as new columns. And the partner wants the flagged privilege rows routed to a specific reviewer before end of day.

Three straightforward changes. Together they take most of an afternoon.

Review Tables already make it easier to pull structured answers from large document sets. But in legal work, the table rarely stays static once those answers land. Productions arrive in waves, privilege calls get challenged, and a partner reframes the question at 4 p.m. The review changes because the matter changes.

With the release of Harvey II in August, the Harvey Agent can take on more complex work across the platform. In Review Tables, that means teams can adapt the table as the matter evolves by delegating any action to the agent in plain language, all from a single chat.

Change a Table by Asking in Plain Language

In the scenario we walked through earlier, each task the litigation team needed to run is now part of a single request.

First, the team asks Harvey to make privilege designation a classification and re-run that column, so a column that came back in the wrong shape is fixed in place rather than rebuilt.

Then, they attach the Relativity export and ask for the custodian metadata as new columns, so custodian and privilege sit in one table and can be filtered together.

Finally, they ask Harvey to flag the privileged rows in red and assign them to the reviewer, so the partner has a routed table before end of day.

These capabilities extend beyond privilege review. Teams can revise and rerun a column’s prompt, output format, or model without rebuilding it. They can also apply the same edit across hundreds of cells, add rows as new documents arrive, or remove rows based on specific criteria.

Data from an iManage load file, CSV, or web search can also be added directly to columns, keeping it in the review table instead of a separate spreadsheet as seen in this video.

For some reviews, related documents need to be considered together rather than as separate rows. Ask Harvey to group email threads with their attachments or complaints with their exhibits, like in the video below. Harvey identifies these relationships using both the source files and the extracted table content, allowing reviewers to assess each complete thread or filing without linking the documents manually. You can see that happening in the following video.

Find a Review Table by Describing it to Harvey

Before Harvey can do any of the actions outlined above, you have to find the right Review Table. Harvey now lets you retrieve a review table through an open-ended query instead of searching for it manually.

Ask, "Find my most recent review table" or describe what the table covered, and the agent locates and attaches it so you can continue querying or editing. For teams with multiple active reviews across different matters, this removes a consistent friction point: you no longer need to remember where a table lives or navigate to it directly before asking a question about it.

From Review Table to Deliverable in One Chat

The most common workflow after building a Review Table is using it to create something else, like a memo, summary, or chart. That handoff previously required switching threads: finishing the table, then starting a new thread and referencing what you just built.

Now, you can ask Harvey to extract information across your documents into a review table, then immediately build a memo or visualization on top of it, all within the same chat.

Putting it to Work

These capabilities are built for the teams doing the highest-volume document work: M&A and diligence teams processing large contract portfolios, and litigation teams reviewing discovery productions.

The agent operates with the same permissions as the user. Existing Vault access controls and ethical walls apply exactly as they do today. A Review Table should be able to change as quickly as the matter does. With the Harvey Agent, teams can keep the review moving without rebuilding the work every time something changes.

Agentic Review Table actions are generally available now. Reach out to your Harvey representative to learn more, or request a demo below if you aren't yet a Harvey customer.

This month's releases let you personalize Harvey to your preferred style, accelerate critical document reviews, and tap into broader context across your systems and knowledge sources.

Tailor Harvey to How You Work

Harvey now remembers how you work and helps you embed team branding into reusable resources.

Personalize Harvey With User-Level Memory

Tell Harvey once to remember a preference, such as your drafting style or your usual negotiating posture, and apply those memories automatically across the platform.

Learn more.

Template Library

Automatically generate documents, client decks, and memos in your team's branded format, and save those templates to scale consistency across outputs.

Learn more.

Custom Agents in Word

Produce consistent results across agreements and contracts by running custom agents directly in Word.

Learn more.

Delegate Large-Scale Document Review to Agents

Speed up due diligence reviews and evidence analysis with better ways to build, search, and edit review tables.

Agentic Vault Search

Ask a question across an entire data room or corpus of files by attaching the full Vault or working within a Space. Harvey finds the relevant files itself and cites the specific pages behind its answer.

Learn more.

Review Table Agent Actions

Restructure a review table by describing what you want, whether that's bulk editing cells, auto-grouping rows, or importing document metadata, and Harvey makes the change directly.

Learn more.

New Review Table Creation Flow

Describe the columns you need in plain language and Harvey builds the table with you, instead of starting from an empty grid.

Learn more.

Review Table Homepage

Find any table you've built or that a colleague has shared with you from a single Review Tables page, with filters, sorting, and search.

Learn more.

Ground Research in Authoritative Sources

Tap into a growing set of trusted legal data sources, and include more sources at once for more comprehensive analysis.

Enhanced Multi-Source Queries

Answer a cross-border question in one pass by attaching up to five knowledge sources in a single query.

Learn more.

150+ New Legal Research Sources

Research against more than 150 new authoritative sources added this month, including content from Chile Tirant lo Blanch. See the sources added here, here, and here.

Move Through Redlines With Fewer Clicks

Streamline negotiations and contract reviews in Harvey for Word.

Alternate Suggestions in the Harvey for Word Add-in

Compare Harvey's options for a clause side-by-side on the suggestion card, then apply the wording that fits your position.

Learn more.

Apply Suggestions From History

Pick up a draft you started earlier by opening the thread from history and applying the suggestions you passed over the first time.

Learn more.

Keep Work Connected and Secure Across Your Existing Tools

Search across your team’s existing precedent with greater ease, and keep roles and work tied to the right permissions.

iManage Search

Find precedent by describing what you need rather than hunting for the file; Harvey searches your iManage content and answers with citations back to the source documents, in either the web app or the Word Add-in.

Learn more.

Client Matter Enhancements

Assign a client matter to a Playbook, search matters by name or description instead of number, and require lawyers to acknowledge a client's AI policy before they work on the matter.

Learn more.

User Attribute Sync via SCIM

Keep user profiles accurate without manual upkeep by automatically syncing job title, practice area, and location from Entra ID or Okta, allowing you to get sharper adoption data in Command Center.

Learn more.

That's all for September. Try one of these features today, or reach out to your Harvey team to learn more. If you're not a Harvey customer, get in touch for a demo.

Turn your team’s precedents, guidelines, and judgment into a custom review agent that accelerates reviews and improves negotiated outcomes.

Every contract that crosses an in-house team's desk carries the weight of decisions that came before it: the fallback accepted last quarter, the clause that always escalates to a commercial approver, the payment terms that got the last three MSAs signed. That accumulated judgment is valuable institutional knowledge, yet much of it remains scattered.

Over the past year, we've worked closely with in-house legal teams of all sizes to understand how contract review actually happens. Playbooks work well for high-volume, standardized agreements, but not every review fits neatly within them. They can capture a team’s stated positions, but not always the nuance behind them—from acceptable fallback ranges to patterns in past negotiations that show what the team will actually accept.

We believe a legal team's negotiation judgment is institutional knowledge worth capturing and scaling. That’s why we built Contract Review Agents, the next evolution of contract review in Harvey, and a key step toward our Contract Intelligence vision, so the business can move faster and get better negotiated outcomes. Instead of a static set of rules, legal teams can create dynamic agents from their precedents, guidelines, and playbooks, plus the judgment that never made it onto paper. Every contract is reviewed by an agent that applies your standards to the redlines, and every review sharpens the agent’s recommendations for the next one.

Build on the Knowledge You Already Have

A Contract Review Agent starts with the knowledge your team already relies on, including executed contracts, negotiation guidelines, and playbooks, to establish how it should review.

Then, Harvey helps capture what is not written down. Through a guided, agent-led interview, Harvey asks targeted questions about how your team applies its standards in practice, including:

• The liability cap your team proposes and the fallback it will accept
• How positions change based on the counterparty, deal value, or business context
• Which clauses always escalate and which the team resolves routinely
• What past deals reveal about the terms the team actually accepts

Together, these sources and answers give Harvey a durable understanding of how your team reviews and negotiates. Harvey turns that institutional knowledge into flexible agents for NDAs, MSAs, DPAs, and other contract types. With Memory, every Harvey agent also remembers individual lawyer preferences, like a light touch on tracked changes, so reviews reflect both your team’s standards and how each person works.

Benchmark Every Redline Against Past Deals

Every new contract benefits from what your team learned negotiating the ones before it. When a redlined MSA lands, the agent knows which changes align with your standards, where to push back, and what needs escalation. It delivers a first-pass review with recommended edits, reasoning, and citations to the sources behind each decision.

The harder moment comes when the counterparty pushes back and the question becomes: What have we actually agreed to before?

Answering this typically means relying on memory or searching through old agreements. With precedent benchmarking, Contract Review Agents can search through an attached precedent bank to bring that answer directly into the review. Alongside each redline, you can see how the counterparty’s proposal compares with positions your team has accepted in executed contracts, visualized clause by clause and cited back to the source documents.

Keep Your Review Standards Current

Negotiation positions change. A playbook written six months ago no longer reflects what the team is agreeing to today. Contract Review Agents can use completed deal data to identify where negotiated outcomes are starting to diverge from the agent’s current standards.

Harvey can parse through this data to surface inconsistencies in what your team is agreeing to, flag when an agent’s positions have fallen behind recent negotiations, and recommend specific updates for the team to review. If the last five NDAs all accepted a term the agent continues to flag, for example, Harvey recognizes the pattern and proactively recommends a new standard, keeping future reviews current and focused on the changes that will drive better outcomes.

Your team's negotiation expertise took years to build. Contract Review Agents make that expertise available on every review, grounding new decisions in the standards and precedents your team has already established. To build your own Contract Review Agent, reach out to your Harvey account team to join the early access waitlist.

Turn your team’s precedents, guidelines, and judgment into a custom review agent that accelerates reviews and improves negotiated outcomes.

Every contract that crosses an in-house team's desk carries the weight of decisions that came before it: the fallback accepted last quarter, the clause that always escalates to a commercial approver, the payment terms that got the last three MSAs signed. That accumulated judgment is valuable institutional knowledge, yet much of it remains scattered.

Over the past year, we've worked closely with in-house legal teams of all sizes to understand how contract review actually happens. Playbooks work well for high-volume, standardized agreements, but not every review fits neatly within them. They can capture a team’s stated positions, but not always the nuance behind them—from acceptable fallback ranges to patterns in past negotiations that show what the team will actually accept.

We believe a legal team's negotiation judgment is institutional knowledge worth capturing and scaling. That’s why we built Contract Review Agents, the next evolution of contract review in Harvey, and a key step toward our Contract Intelligence vision, so the business can move faster and get better negotiated outcomes. Instead of a static set of rules, legal teams can create dynamic agents from their precedents, guidelines, and playbooks, plus the judgment that never made it onto paper. Every contract is reviewed by an agent that applies your standards to the redlines, and every review sharpens the agent’s recommendations for the next one.

Build on the Knowledge You Already Have

A Contract Review Agent starts with the knowledge your team already relies on, including executed contracts, negotiation guidelines, and playbooks, to establish how it should review.

Then, Harvey helps capture what is not written down. Through a guided, agent-led interview, Harvey asks targeted questions about how your team applies its standards in practice, including:

• The liability cap your team proposes and the fallback it will accept
• How positions change based on the counterparty, deal value, or business context
• Which clauses always escalate and which the team resolves routinely
• What past deals reveal about the terms the team actually accepts

Together, these sources and answers give Harvey a durable understanding of how your team reviews and negotiates. Harvey turns that institutional knowledge into flexible agents for NDAs, MSAs, DPAs, and other contract types. With Memory, every Harvey agent also remembers individual lawyer preferences, like a light touch on tracked changes, so reviews reflect both your team’s standards and how each person works.

Benchmark Every Redline Against Past Deals

Every new contract benefits from what your team learned negotiating the ones before it. When a redlined MSA lands, the agent knows which changes align with your standards, where to push back, and what needs escalation. It delivers a first-pass review with recommended edits, reasoning, and citations to the sources behind each decision.

The harder moment comes when the counterparty pushes back and the question becomes:

What have we actually agreed to before?

Answering this typically means relying on memory or searching through old agreements. With precedent benchmarking, Contract Review Agents can search through an attached precedent bank to bring that answer directly into the review. Alongside each redline, you can see how the counterparty’s proposal compares with positions your team has accepted in executed contracts, visualized clause by clause and cited back to the source documents.

Keep Your Review Standards Current

Negotiation positions change. A playbook written six months ago no longer reflects what the team is agreeing to today. Contract Review Agents can use completed deal data to identify where negotiated outcomes are starting to diverge from the agent’s current standards.

Harvey can parse through this data to surface inconsistencies in what your team is agreeing to, flag when an agent’s positions have fallen behind recent negotiations, and recommend specific updates for the team to review. If the last five NDAs all accepted a term the agent continues to flag, for example, Harvey recognizes the pattern and proactively recommends a new standard, keeping future reviews current and focused on the changes that will drive better outcomes.

Your team's negotiation expertise took years to build. Contract Review Agents make that expertise available on every review, grounding new decisions in the standards and precedents your team has already established. To build your own Contract Review Agent, reach out to your Harvey account team to join the early access waitlist.

A partner asks for every change of control provision across the vendor agreements. Before anyone can answer, someone has to first work out which files in a folder named "Contracts_Round2_FINAL" actually contain those clauses. Sometimes that's a quick scan. Sometimes it's an associate spending an afternoon opening file after file to find the handful that matter.

AI already helps with the second half of this problem. Hand any model the right twelve agreements and it will pull every change of control provision and flag the ones with odd carve-outs. The issue? Someone still had to find the right twelve agreements. Keyword searching and folder structure isn’t usually consistent enough to make this task an easy one.

We’re announcing two new features aimed to help solve this problem: Agentic Vault Search and Agentic Vault Organization. In Harvey, Vault is where legal teams store and query large document sets, including data rooms, discovery productions, and contract portfolios.

The Harvey Agent Finds Documents Without Needing to be Told

Most AI search tools are still doing keyword matching underneath, just a fuzzier version that gets you closer when your words don't line up exactly with the document's. The Harvey Agent works differently, looking at which folder a file sits in, what the client's naming convention tends to mean, and who uploaded it — the same signals an associate would use when going through a physical data room. You can see this portrayed in the following video.

With this release, you don't have to tell the Harvey Agent which folder to check or what a document is called. Attach a full Vault to your query, or work inside a Space that already has vaults attached, and the agent finds the right document on its own.

The agent has the same permissions as the user: ethical walls and existing Vault permissions (view, edit, full access) apply exactly as they do today, and every file the agent touches is logged in the workspace's audit trail.

The Harvey Agent Can Organize Files, not Just Locate Them

Most incoming document sets are messy before anyone even starts searching through them. Counterparts hand over data rooms with no consistent naming convention. Discovery productions arrive as a folder of files named using nothing but numbers.

Agentic Vault Search finds you the right files when you ask for them. File organization is a different problem, since a paralegal browsing the folder directly, a partner exporting the Vault for a closing binder, or a client looking at their own data room never types a query at all, and for them the underlying structure is the only organization they can rely on.

With Agentic Vault Organization, tell Harvey in plain language what you want done, such as "organize this vault into folders by counterparty and agreement type," or "rename everything in the Contracts folder to follow [Counterparty] - [Agreement Type] - [Date]." Harvey proposes a plan and shows exactly what it intends to change before touching anything, so you confirm or adjust before it executes. You can also undo the most recent set of changes just in case.

The Harvey Agent can move, rename, and tag files, but importantly, it never edits the content inside them and never runs a change without a person confirming it first.

Putting Agentic Vault to Work

Agentic Vault Search and Organization is built for M&A and diligence teams working inside large data rooms, litigation teams working through discovery productions, and any practice group managing a Vault whose size or complexity makes it difficult to navigate.

Let Harvey do the tedious work of renaming or organizing your files so you can have more time to focus on the work that matters most.

Agentic Vault Search is generally available now. If you are interested in Agentic Vault Organization, reach out to your Harvey representative to be added to the Early Access program.

If you aren’t a Harvey customer, reach out to our team below for a demo:

Agentic Vault Search and Organization help legal teams find the right documents and bring more structure to large, complex Vaults.

A partner asks for every change of control provision across the vendor agreements. Before anyone can answer, someone has to first work out which files in a folder named "Contracts_Round2_FINAL" actually contain those clauses. Sometimes that's a quick scan. Sometimes it's an associate spending an afternoon opening file after file to find the handful that matter.

AI already helps with the second half of this problem. Hand any model the right twelve agreements and it will pull every change of control provision and flag the ones with odd carve-outs. The issue? Someone still had to find the right twelve agreements. Keyword searching and folder structure isn’t usually consistent enough to make this task an easy one.

We’re announcing two new features aimed to help solve this problem: Agentic Vault Search and Agentic Vault Organization. In Harvey, Vault is where legal teams store and query large document sets, including data rooms, discovery productions, and contract portfolios.

The Harvey Agent Finds Documents Without Needing to be Told

Most AI search tools are still doing keyword matching underneath, just a fuzzier version that gets you closer when your words don't line up exactly with the document's. The Harvey Agent works differently, looking at which folder a file sits in, what the client's naming convention tends to mean, and who uploaded it — the same signals an associate would use when going through a physical data room. You can see this portrayed in the following video.

With this release, you don't have to tell the Harvey Agent which folder to check or what a document is called. Attach a full Vault to your query, or work inside a Space that already has vaults attached, and the agent finds the right document on its own.

The agent has the same permissions as the user: ethical walls and existing Vault permissions (view, edit, full access) apply exactly as they do today, and every file the agent touches is logged in the workspace's audit trail.

The Harvey Agent Can Organize Files, not Just Locate Them

Most incoming document sets are messy before anyone even starts searching through them. Counterparts hand over data rooms with no consistent naming convention. Discovery productions arrive as a folder of files named using nothing but numbers.

Agentic Vault Search finds you the right files when you ask for them. File organization is a different problem, since a paralegal browsing the folder directly, a partner exporting the Vault for a closing binder, or a client looking at their own data room never types a query at all, and for them the underlying structure is the only organization they can rely on.

With Agentic Vault Organization, tell Harvey in plain language what you want done, such as "organize this vault into folders by counterparty and agreement type," or "rename everything in the Contracts folder to follow [Counterparty] - [Agreement Type] - [Date]." Harvey proposes a plan and shows exactly what it intends to change before touching anything, so you confirm or adjust before it executes. You can also undo the most recent set of changes just in case.

The Harvey Agent can move, rename, and tag files, but importantly, it never edits the content inside them and never runs a change without a person confirming it first.

Putting Agentic Vault to Work

Agentic Vault Search and Organization is built for M&A and diligence teams working inside large data rooms, litigation teams working through discovery productions, and any practice group managing a Vault whose size or complexity makes it difficult to navigate.

Let Harvey do the tedious work of renaming or organizing your files so you can have more time to focus on the work that matters most.

Agentic Vault Search is generally available now. If you are interested in Agentic Vault Organization, reach out to your Harvey representative to be added to the Early Access program.

If you aren’t a Harvey customer, reach out to our team below for a demo:

Announcing Fable 5.1 in Harvey

Claude Fable 5.1, the latest Anthropic model, is now available. Starting today, Harvey customers can opt-in to receive access.

In our early testing, Fable 5.1 was strongest on complex, document-heavy analysis, with the most significant quality improvement on litigation and dispute resolution compared to Fable 5. Generally, Fable 5.1 also performed well across tasks related to data privacy and cybersecurity and capital markets.

Notably, Fable 5.1 also delivers quality results even at lower reasoning levels. At its lowest reasoning setting, Fable 5.1 was thorough in completing tasks, while using roughly half the output tokens. Balancing quality and overall efficiency is meaningful, especially for legal teams working through high-volume work, such as reviewing agreements.

For full results against our Legal Agent Benchmark, you can refer to the public leaderboards on Artificial Analysis and Vals.

What to Know Before Enabling Fable 5.1

The use of Fable models, including Fable 5.1 and Fable 5, may involve data processing practices that differ from data commitments we made in our existing customer agreements. Eligible Harvey customers can use Fable 5 and Fable 5.1 with zero data retention under a time-bound exemption while Anthropic rolls out Enterprise Frontier Safeguards. Enterprise Frontier Safeguards will give eligible customers the ability to deploy Anthropic's most capable models while storing their data in cloud infrastructure controlled by Harvey or by the customer.

Note Anthropic does not offer any regional processing for Fable 5.1 or Fable 5, so all data will be processed in the United States.

We are making this model available as an opt-in option vs. by default. Our goal is to give customers access to a highly capable model while preserving their ability to decide what’s right for their organization.

Customers who have opted in to Fable 5 will get Fable 5.1 access in Harvey starting this week.

If you have questions or are interested in enabling Fable models for your organization, please reach out to your account team.

Horizon Scanning helps legal teams monitor, assess, and take action on the changes that matter most.

The pace of regulatory change isn't slowing down: 63% of legal departments say their monitoring and compliance workloads increased in 2026. Many existing monitoring tools don't even save teams time because alerts arrive through fragmented feeds, without prioritization, and siloed from where next steps need to happen. Changes can slip through, and it can be costly to miss a key update.

We're introducing Horizon Scanning in Harvey to help legal teams turn monitoring into actionable next steps. Teams can customize alerts, receive notifications for key regulatory and legislative updates, and immediately take action on the most impactful developments in one connected platform. Built with a cohort of design partners across law firms and in-house teams, Horizon Scanning helps legal teams quickly act on the changes that matter most.

Create Tailored Scans Relevant to You

Most monitoring tools draw from a fixed repository, which means coverage may be rigid or irrelevant to what your business prioritizes. With Horizon Scanning in Harvey, teams can simply describe what they need to track in plain language — a topic, a region, a named regulator. Harvey will automatically build a scan, drawing on more than 12,000 sources across 100+ jurisdictions alongside public sources the user describes. For example, an in-house team can shape the scope of a scan to data protection developments in the jurisdictions where the company actually holds data, or create a separate scan for each subsidiary.

Quickly Identify Which Changes Matter Most

Every scan feeds a single dashboard that shows what moved, with the most consequential developments surfaced first. Determining what changed is just the first step — the critical work is understanding who it affects, what they need to do, and by when. Each update arrives with a relevance assessment, so that it is easy to gauge urgency before digging into the details. Teams can also receive email digests of daily changes, helping them access insights directly in their inbox.

Let's say there's a new anti-money laundering rule that impacts a financial services legal team. They can find out about the change the same day it's published via an email alert. Then, Harvey can help them quickly determine which entities are impacted, what has to change operationally, and how long they have to prepare.

Move From Update to Next Step

Since Horizon Scanning is part of Harvey’s agentic platform, alerts connect to downstream work in one single system. Ask a question about a development in plain language and get an answer grounded in the source. Teams can turn scans into a memo revision, a recommended policy update, or an interactive timeline of upcoming obligations without leaving the dashboard.

For a firm's privacy practice, that can mean tracking data protection regulators across member states, updating guidance by directly pulling the right documents from their DMS into queries, and drafting a client alert the same day new guidance lands. For an in-house compliance team, it can mean Harvey directly flagging which internal policies are affected by a tariff change and suggesting revisions, helping them draft new guidelines and flag any operational concerns faster. Scans can be shared across a team, so practice groups and business stakeholders work from the same view.

We're building Horizon Scanning as a foundation for proactive legal work in Harvey, where monitoring, analysis, and the work that follows all live in one place.

Horizon Scanning is currently available in Early Access for Harvey customers, with general availability to follow.

If you’re interested in learning more about Harvey, get in touch with our team.

Announcing Fable 5.1 in Harvey

Claude Fable 5.1, the latest Anthropic model, is now available. Starting today, Harvey customers can opt-in to receive access.

In our early testing, Fable 5.1 was strongest on complex, document-heavy analysis, with the most significant quality improvement on litigation and dispute resolution compared to Fable 5. Generally, Fable 5.1 also performed well across tasks related to data privacy and cybersecurity and capital markets.

Notably, Fable 5.1 also delivers quality results even at lower reasoning levels. At its lowest reasoning setting, Fable 5.1 was thorough in completing tasks, while using roughly half the output tokens. Balancing quality and overall efficiency is meaningful, especially for legal teams working through high-volume work, such as reviewing agreements.

For full results against our Legal Agent Benchmark, you can refer to the public leaderboards on Artificial Analysis and Vals.

What to Know Before Enabling Fable 5.1

The use of Fable models, including Fable 5.1 and Fable 5, may involve data processing practices that differ from data commitments we made in our existing customer agreements. Eligible Harvey customers can use Fable 5 and Fable 5.1 with zero data retention under a time-bound exemption while Anthropic rolls out Enterprise Frontier Safeguards. Enterprise Frontier Safeguards will give eligible customers the ability to deploy Anthropic's most capable models while storing their data in cloud infrastructure controlled by Harvey or by the customer.

Note Anthropic does not offer any regional processing for Fable 5.1 or Fable 5, so all data will be processed in the United States.

We are making this model available as an opt-in option vs. by default. Our goal is to give customers access to a highly capable model while preserving their ability to decide what’s right for their organization.

Customers who have opted in to Fable 5 will get Fable 5.1 access in Harvey starting this week.

If you have questions or are interested in enabling Fable models for your organization, please reach out to your account team.

Horizon Scanning helps legal teams monitor, assess, and take action on the changes that matter most.

The pace of regulatory change isn't slowing down: 63% of legal departments say their monitoring and compliance workloads increased in 2026. Many existing monitoring tools don't even save teams time because alerts arrive through fragmented feeds, without prioritization, and siloed from where next steps need to happen. Changes can slip through, and it can be costly to miss a key update.

We're introducing Horizon Scanning in Harvey to help legal teams turn monitoring into actionable next steps. Teams can customize alerts, receive notifications for key regulatory and legislative updates, and immediately take action on the most impactful developments in one connected platform. Built with a cohort of design partners across law firms and in-house teams, Horizon Scanning helps legal teams quickly act on the changes that matter most.

Create Tailored Scans Relevant to You

Most monitoring tools draw from a fixed repository, which means coverage may be rigid or irrelevant to what your business prioritizes. With Horizon Scanning in Harvey, teams can simply describe what they need to track in plain language — a topic, a region, a named regulator. Harvey will automatically build a scan, drawing on more than 12,000 sources across 100+ jurisdictions alongside public sources the user describes. For example, an in-house team can shape the scope of a scan to data protection developments in the jurisdictions where the company actually holds data, or create a separate scan for each subsidiary.

Quickly Identify Which Changes Matter Most

Every scan feeds a single dashboard that shows what moved, with the most consequential developments surfaced first. Determining what changed is just the first step — the critical work is understanding who it affects, what they need to do, and by when. Each update arrives with a relevance assessment, so that it is easy to gauge urgency before digging into the details. Teams can also receive email digests of daily changes, helping them access insights directly in their inbox.

Let's say there's a new anti-money laundering rule that impacts a financial services legal team. They can find out about the change the same day it's published via an email alert. Then, Harvey can help them quickly determine which entities are impacted, what has to change operationally, and how long they have to prepare.

Move From Update to Next Step

Since Horizon Scanning is part of Harvey’s agentic platform, alerts connect to downstream work in one single system. Ask a question about a development in plain language and get an answer grounded in the source. Teams can turn scans into a memo revision, a recommended policy update, or an interactive timeline of upcoming obligations without leaving the dashboard.

For a firm's privacy practice, that can mean tracking data protection regulators across member states, updating guidance by directly pulling the right documents from their DMS into queries, and drafting a client alert the same day new guidance lands. For an in-house compliance team, it can mean Harvey directly flagging which internal policies are affected by a tariff change and suggesting revisions, helping them draft new guidelines and flag any operational concerns faster. Scans can be shared across a team, so practice groups and business stakeholders work from the same view.

We're building Horizon Scanning as a foundation for proactive legal work in Harvey, where monitoring, analysis, and the work that follows all live in one place.

Horizon Scanning is currently available in Early Access for Harvey customers, with general availability to follow.

If you’re interested in learning more about Harvey, get in touch with our team.

Harvey remembers how you work — carrying your preferences across every session and surface, so each output reflects how you actually practice.

Every lawyer has their non-negotiables. The Oxford comma is essential, or it's clutter. "Shall" is sacred, or it's archaic and struck on sight. Anglo-Saxon rooted words preferred to Romance-rooted words. Until now, AI made you re-teach these preferences in every prompt.

Memory, released as part of Harvey II, changes that. It closes the gap between how you practice and how Harvey executes — automatically, transparently, and always under your control.

Standing Instructions, Not Repeated Prompts

A memory is like a standing instruction that applies itself: Once set, it applies to all your work moving forward. Getting started takes just a few questions about how you work. After that, you can tell Harvey to remember something in the middle of what you're already doing. For example:

• The GC I email reads on his phone. Keep email drafts to him under 150 words, with the ask in the subject line.
• When there is a judgment call, give me two options — a conservative 'safe harbor' path and an 'aggressive' path — each with a risk level.
• Always distinguish the UK GDPR position from the EU GDPR position. Never blend them into a generic “GDPR” answer.
• I am admitted only in New York. Flag anything that requires Delaware or other local counsel input.

Save memories one at a time or capture several at once. The more you tell Harvey, the more it adapts — and the more your work comes back the way you'd do it yourself.

Saved Once, Applied Everywhere

A preference saved once travels with you across the Web App, Outlook, and the Word Add-In, in playbook runs, and every agent you put to work. Tell Harvey how you like your partner or GC updates written. Harvey maintains those standards everywhere you draft updates, whether in a doc or an email, with no reminder and no reformatting.

This is what makes agents more useful, not just faster. An agent that carries your context does every step of the work your way: your framing, your positions, your format. Run a dozen agents at once, and each one will work your way. That means more work completed, all to your standard, giving you time back for higher-value work.

And wherever a memory is applied, you can see it. When a memory shapes a response, Harvey cites it in it the same way it cites a source. You can see which preferences were used, confirm it’s the right call, and catch any instances that need correction.

You Control What Harvey Remembers

General-purpose AI assistants treat memory as a background process: the model quietly accumulates what it learns about you, and you find out what it stored, or how it's being used, only if you go digging. That's fine for consumer chat. It's below the bar for legal work.

Harvey is built on the opposite principle: you always know what it remembers, when it learned it, and when it's being used.

Harvey will save memories when you explicitly ask (“Remember to…”) and when a prompt signals a preference (“I want all summaries to start with BLUF”). In both cases, a confirmation card shows exactly what was saved so you can control what Harvey stores. Every memory lives in one place in a user’s Personal Settings, where they can review, edit, search, or delete any memories, or turn Memory off entirely.

Admins control whether Memory is available in their workspace, including whether to enable it for all users or a select subset. And it's protected by the same security protections as everything else in Harvey. Memory opens no new access to your systems, and like the rest of a customer’s workspace content in Harvey, individual memories are never used to train AI models.

Make Harvey Yours

:Harvey:'s Memory feature lets me spend less time re-establishing context and more time doing sophisticated work. It remembers my role, drafting style, analytical approach, and output preferences, so I can focus on better prompts, deeper follow-up questions, and higher-quality legal analysis.

• Sam Sadighi, VP, Legal

An organization’s institutional context lives with its people: their expertise, insights and understanding of what matters most to their business and customers. We have worked with :Harvey: to shape Memory, which carries this growing context forward so users can build on it each time they use the tool. Critically, this context is stored securely, appropriately governed and being used in the right way.

• Jessica Escalera, Managing Director, Chief AI Officer, Legal

Working with :Harvey: on Memory early on, we emphasized that users need full transparency and control over what is stored and used — a non-negotiable requirement for legal work. It’s great to see that principle reflected in how :Harvey: has designed Memory. We're looking forward to seeing our lawyers spend less time repeating context and get closer to a final draft from the first prompt.

• Pierre G. Zickert, Counsel

We actively partnered with :Harvey: on Memory because it enables our people to capture their preferences, standards and drafting approach once, then apply them consistently across matters and engagements. One of the most powerful unlocks is in its ability to understand how we each work, not just respond to what we ask. Personalisation like this makes AI feel like a genuine extension of our lawyers, rather than just another tool.

• Michelle Mahoney, Chief Innovation Officer

Since announcing Memory in January, we've shaped it with the teams who will rely on it — AmLaw and Magic Circle firms, Fortune 500 and FTSE legal departments — across the US, UK, Europe, the Middle East, and Asia-Pacific.

Memory today is scoped to you alone: your preferences, in your workspace, visible to no one else without your knowledge. It's our introduction to memory as a concept in Harvey, and it's the foundation for what comes next. We're already building memory scoped to a shared space, so context will stay tied to the right client matter or project, with organization-level memory to follow.

Memory is available in Early Access today, with general availability in the next couple of weeks. Reach out to your Harvey account team or contact us for a demo.

Harvey remembers how you work — carrying your preferences across every session and surface, so each output reflects how you actually practice.

Every lawyer has their non-negotiables. The Oxford comma is essential, or it's clutter. "Shall" is sacred, or it's archaic and struck on sight. Anglo-Saxon rooted words preferred to Romance-rooted words. Until now, AI made you re-teach these preferences in every prompt.

Memory, released as part of Harvey II, changes that. It closes the gap between how you practice and how Harvey executes — automatically, transparently, and always under your control.

Standing Instructions, Not Repeated Prompts

A memory is like a standing instruction that applies itself: Once set, it applies to all your work moving forward. Getting started takes just a few questions about how you work. After that, you can tell Harvey to remember something in the middle of what you're already doing. For example:

• The GC I email reads on his phone. Keep email drafts to him under 150 words, with the ask in the subject line.
• When there is a judgment call, give me two options — a conservative 'safe harbor' path and an 'aggressive' path — each with a risk level.
• Always distinguish the UK GDPR position from the EU GDPR position. Never blend them into a generic “GDPR” answer.
• I am admitted only in New York. Flag anything that requires Delaware or other local counsel input.

Save memories one at a time or capture several at once. The more you tell Harvey, the more it adapts — and the more your work comes back the way you'd do it yourself.

Saved Once, Applied Everywhere

A preference saved once travels with you across the Web App, Outlook, and the Word Add-In, in playbook runs, and every agent you put to work. Tell Harvey how you like your partner or GC updates written. Harvey maintains those standards everywhere you draft updates, whether in a doc or an email, with no reminder and no reformatting.

This is what makes agents more useful, not just faster. An agent that carries your context does every step of the work your way: your framing, your positions, your format. Run a dozen agents at once, and each one will work your way. That means more work completed, all to your standard, giving you time back for higher-value work.

And wherever a memory is applied, you can see it. When a memory shapes a response, Harvey cites it in it the same way it cites a source. You can see which preferences were used, confirm it’s the right call, and catch any instances that need correction.

You Control What Harvey Remembers

General-purpose AI assistants treat memory as a background process: the model quietly accumulates what it learns about you, and you find out what it stored, or how it's being used, only if you go digging. That's fine for consumer chat. It's below the bar for legal work.

Harvey is built on the opposite principle: you always know what it remembers, when it learned it, and when it's being used.

Harvey will save memories when you explicitly ask (“Remember to…”) and when a prompt signals a preference (“I want all summaries to start with BLUF”). In both cases, a confirmation card shows exactly what was saved so you can control what Harvey stores. Every memory lives in one place in a user’s Personal Settings, where they can review, edit, search, or delete any memories, or turn Memory off entirely.

Admins control whether Memory is available in their workspace, including whether to enable it for all users or a select subset. And it's protected by the same security protections as everything else in Harvey. Memory opens no new access to your systems, and like the rest of a customer’s workspace content in Harvey, individual memories are never used to train AI models.

Make Harvey Yours

:Harvey:'s Memory feature lets me spend less time re-establishing context and more time doing sophisticated work. It remembers my role, drafting style, analytical approach, and output preferences, so I can focus on better prompts, deeper follow-up questions, and higher-quality legal analysis. — Sam Sadighi, VP, Legal

An organization’s institutional context lives with its people: their expertise, insights and understanding of what matters most to their business and customers. We have worked with :Harvey: to shape Memory, which carries this growing context forward so users can build on it each time they use the tool. Critically, this context is stored securely, appropriately governed and being used in the right way. — Jessica Escalera, Managing Director, Chief AI Officer, Legal

Working with :Harvey: on Memory early on, we emphasized that users need full transparency and control over what is stored and used — a non-negotiable requirement for legal work. It’s great to see that principle reflected in how :Harvey: has designed Memory. We're looking forward to seeing our lawyers spend less time repeating context and get closer to a final draft from the first prompt. — Pierre G. Zickert, Counsel

We actively partnered with :Harvey: on Memory because it enables our people to capture their preferences, standards and drafting approach once, then apply them consistently across matters and engagements. One of the most powerful unlocks is in its ability to understand how we each work, not just respond to what we ask. Personalisation like this makes AI feel like a genuine extension of our lawyers, rather than just another tool. — Michelle Mahoney, Chief Innovation Officer

Since announcing Memory in January, we've shaped it with the teams who will rely on it — AmLaw and Magic Circle firms, Fortune 500 and FTSE legal departments — across the US, UK, Europe, the Middle East, and Asia-Pacific.

Memory today is scoped to you alone: your preferences, in your workspace, visible to no one else without your knowledge. It's our introduction to memory as a concept in Harvey, and it's the foundation for what comes next. We're already building memory scoped to a shared space, so context will stay tied to the right client matter or project, with organization-level memory to follow.

Memory is available in Early Access today, with general availability in the next couple of weeks. Reach out to your Harvey account team or contact us for a demo.