Regional Knowledge Sources Overview

A comprehensive overview of Harvey's integrated public data sources for legislation, case law, and regulatory materials, including current availability and planned rollout timelines.

Last updated: Feb 12, 2026


Overview

Harvey's Regional Knowledge Sources let you ground answers to your queries in curated, jurisdiction-specific legal materials such as legislation and case law.

Harvey’s regional Knowledge Sources allow you to query authoritative, publicly available legal and regulatory materials directly within Harvey. These sources help ground responses in jurisdiction-specific legislation, case law, and regulatory guidance, with precise citations.

To jump to a full list of these sources, click Harvey's Regional Knowledge Sources in the Table of Contents (right-side menu).


How Harvey Retrieves Content for Regional Knowledge Sources

For Regional Knowledge Sources, Harvey selects the legal resources that may be referenced for each jurisdiction. These typically include official legislation sites, court websites, and regulator portals that lawyers would already treat as authoritative.

When users select a Regional Knowledge Source during a query, Harvey limits its search and citations to a defined set of legal and regulatory sites for that jurisdiction. This helps ensure that responses are based on authoritative primary sources with clear, verifiable citations.

To access these materials and ground user queries, Harvey may employ a variety of methods to responsibly access accurate information with citations back to the original source. These methods include:

  • Proprietary retrieval workflows
  • Real-time access
  • Curated search queries using Harvey partners approved by our customers (e.g., OpenAI, Parallel).

Knowledge Source Partners

For web-search–based Knowledge Sources, Harvey currently uses OpenAI Web Search. Starting February 2, 2026, Harvey may also leverage Parallel as a sub-processor to support enhanced and expanded access to certain web-based Knowledge Sources.

By default, Parallel is turned off for all Harvey workspaces. Harvey will not send any queries to Parallel unless a workspace admin explicitly opts in to using Parallel for their organization using their Admin Settings. Until that opt‑in occurs, Regional Knowledge Sources will continue to rely on Harvey’s proprietary and partner-based workflows.

Use of specific web search partners is controlled by workspace admins. Harvey will not send any queries to these partners unless a workspace admin explicitly opts in for their organization using their Admin Settings. Until that opt‑in occurs, Regional Knowledge Sources will continue to rely on Harvey’s proprietary and partner-based workflows.

Expected Data Flows

To elaborate on the summary above, Harvey's Regional Knowledge Source data is either pulled from Harvey’s curated internal knowledge-source through Retrieval Augmented Generation (RAG) or with support from secure search APIs enabled Harvey subprocessors (e.g. Open AI, Parallel). Below are the expected data flows in each scenario. Note that Harvey subprocessors are contractually committed to delete all data after processing Harvey user requests.

Image of expected data flows for knowledge sources

Common Use Cases

Click the expanders for ideas on how to leverage regional sources in Harvey.

Data Handling

When users submit query that leverages a Knowledge Source powered by OpenAI or Parallel:

  • Harvey reviews the user’s request and creates a generalized keyword search query for secure transmission to the enabled Knowledge Source provider. By design, keyword search queries shared with Knowledge Source providers do not include customer documents stored in Vault, verbatim prompt text, or sensitive or identifying information such content could contain.
  • All data shared with Knowledge Source providers is encrypted, processed temporarily, and deleted immediately after results are returned.

Harvey requires its Knowledge Source partners (e.g., OpenAI, Parallel) to comply with strict contractual and technical safeguards:

  • No data storage, training, or human review: Neither Parallel nor OpenAI store, train on, or allow human access to Harvey customer data.
  • Temporary processing only: All Knowledge Source queries are processed briefly our partners, and any query-related content is deleted immediately after results are returned.
  • Secure environment: For any query-related content shared with Harvey’s partners, processing occurs within our web processor's secure cloud infrastructures. To learn more about the security programs of Harvey’s web search providers, see here for Parallel and here for OpenAI.
  • End-to-end encryption (TLS 1.2+ in transit, AES-256+ at rest)

Important: If enabled, any customer materials that must be shared with Parallel are processed by Parallel exclusively within the United States regardless of your workspace’s regional hosting set up.

For additional details, see Harvey’s Service Terms and Trust Center.


Add a Knowledge Source to Your Prompt

Recommended sources will populate below the prompt window, or you can browse sources by following the steps below.

  1. From within the prompt window, click Files and sources to open the full list of sources available to you.
  2. Scroll through the list of sources and select up to two.
  3. From the selected source, you can choose the specific data sources you’d like to use.
  4. Click Add to confirm your selections.
  5. Write your query in the prompt window then Ask Harvey.
    Image of Australia knowledge sources selected
  6. Review Harvey’s response and citations.
    1. In Harvey’s response, hover over a citation to preview the data source that informed the response. You can click View Reference to open the full context of the source.
      Image of hovering over citation in harvey's response
    2. The right-side panel lists an overview of all cited sources. You can close this panel by clicking the X in the top right.
  7. Ask Harvey follow-up questions as needed in the thread.

Harvey's Regional Knowledge Sources

This list reflects currently supported integrations with publicly available data sources covering legislation, case law and other regulatory information.

Early Access (EA) vs. General Availability (GA)

We currently offer regional knowledge sources in both EA (beta) and GA stages. EA sources are fully functional but may still be refined as we actively monitor data quality, reliability, and incorporate user feedback. Once a source consistently meets our standards for stability and performance, it is promoted to GA and becomes fully supported for everyday use.