Twenty-four hours for a price and a conflict clearance.
Forty-eight hours for delivery.
Seventy-two for complex filings over twenty pages.

Powered by Veracity Engine

The quality of your legal research should not depend on the size of your firm. Leveling the field.

DISMANTLE

We take your opponent's filing apart. Every citation checked against the actual source, every quotation compared word for word, every proposition of law tested against the holding it claims to support. You receive a report identifying every deficiency, mischaracterization, phantom citation, and unsupported proposition in the filing.

BUILD (attorneys only)

We draft your responsive filing from the ground up, on verified authority.

ARM (research only, no drafting)

We deliver your research arsenal. Verified authorities, verbatim quotes with pin cites, organized by issue and ready to use. You build the brief; we hand you the materials.

Two-phase flat-fee structure. Phase one: $100 deposit on submission. Phase two: flat fee quoted within 24 hours, individually priced per engagement, deposit applied to the total. If a conflict is identified during the 24-hour window, the deposit is refunded in full and the engagement is declined. Each submission is a discrete engagement with a discrete price. Revisions or follow-on work require a new submission.

Veracity Engine, the verification architecture behind the work, scores 100/0/0 on the Stanford Legal AI benchmark.

In more than three-quarters of state civil cases, at least one party appears without counsel, almost always because counsel is financially unreachable. Level-Field delivers verified, cite-ready legal research to attorneys and self-represented litigants alike.

Attorney — for licensed counsel.
Layperson — for self-represented parties.
Dismantle — tear apart their filing.
Build — draft your responsive filing.
Arm — verified research arsenal, no drafting.

Your $100 deposit is fully refundable if we identify a conflict or decline the engagement.

Level-Field is not a law firm. Submission of materials through this site does not create an attorney-client relationship between you and Level-Field. The supervising attorney of record remains solely responsible for all filings, including independent verification of all work product. Each submission constitutes a discrete engagement.

Other legal AI fails the privilege test.

Heppner set the standard.

In February 2026, the Southern District of New York held in United States v. Heppner that attorney-client privilege fails when an AI platform's terms permit the operator to access the user's data. The court was explicit: it is the capability to access, not the fact of access, that defeats confidentiality.

Kovel preserves one path for AI as a privileged agent of counsel, but only where the operator structurally cannot read the material it processes. A court ruling on that question will look first at the platform's published terms.

The same week, the Eastern District of Michigan reached a different result on similar facts in Warner v. Gilbarco, but only under the work-product doctrine, which waives on disclosure to an adversary. The privilege analysis in Heppner is untouched. A filer planning beyond the Second Circuit should plan against the more demanding standard.

Harvey
Service Terms · last updated April 10, 2026
§4.4"Knowledge Source and Web Browsing subprocessors are not HIPAA compliant."
§4.4 / §4.5"Customer data may not process in Your selected data processing region."
Each disclosed subprocessor is a separate access point under the Heppner analysis.
CoCounsel · Thomson Reuters
Product Specific Terms v2.1 · March 4, 2025
§2.1"[User] authoriz[es] us to access and share Your Data with the third-party provider."
§3.3"the Syncly DMS service may be hosted in a single or multi-tenant environment in our discretion."
Express contractual authorization of third-party disclosure. The operator's discretion is the test, not the customer's preference.
Lexis+ AI · Westlaw Precision AI
The foundation-model gap
Most legal-AI products run on third-party foundation models (OpenAI's GPT family, Anthropic Claude, others). The customer's contract is with the vendor; the inference itself executes on the foundation provider's servers. Zero-data-retention promises in the vendor contract do not bind the upstream provider. Under the Heppner framework, the customer has no privity with the entity that holds the access capability.
ChatGPT · Claude · Gemini
Consumer privacy policies
Heppner, slip op. at 6"[T]he written privacy policy to which users of Claude consent provides that Anthropic … reserves the right to disclose such data to a host of 'third parties,' including 'governmental regulatory authorities.'"
The Heppner holding itself, applied to Anthropic. The ChatGPT and Gemini consumer terms read similarly.

Level-Field's verification work runs on Veracity Engine. Veracity Engine was designed against the Heppner standard: the operator has no path to read the content of a submission. Where Harvey relies on a disclosed subprocessor chain, where CoCounsel authorizes third-party sharing, where Lexis+ AI and Westlaw Precision AI route through foundation models the vendor cannot bind, the pipeline here closes the gap by architecture.

The guarantee is hardware-enforced. Operator access to a submission's content is structurally impossible. Subprocessors with access to substance: zero. Retention: cryptographically attested to zero. Privilege is preserved by the architecture, not by the contract.

  • Files at rest Encrypted in access-restricted infrastructure, available only to Level-Field personnel performing the engagement.
  • Inference path Sovereign Shield isolates the AI provider from the submission content. No access, no retention, attested.
  • Managed Private Inference TEE-backed managed inference. Zero platform retention.
  • Subprocessor scope Zero for content. Billing, hosting, and email-delivery vendors do not see the substance of what was submitted.

Comparison current as of June 2026. Vendor terms can and do change. The method shown here — direct quotation of a specific dated clause — is the same method a court applies in a privilege challenge.