Parambil
Parambil is an AI platform for reading medical records at litigation scale. It ingests unstructured charts that commonly run to thousands of pages across multiple facilities and produces sortable medical chronologies with provider notes, medications, procedures, lab results and imaging, every entry hyperlinked back to the source page. Around that sit a Provider Guide that de-duplicates every clinician in the record and organises them by specialty for deposition preparation, billing analysis, and a missing-data view that flags gaps and documents referenced but absent from the file. On top of the chronology the platform runs agents: a Case Agent that investigates a case end to end, a Drafting Agent that produces first drafts of complaints, demands, interrogatories and discovery requests from case data and a firm's own exemplars, and a live web search that pulls clinical guidelines and medical literature. The agent layer is built on a stated partnership with Anthropic and the web search on a stated partnership with Perplexity. Outputs export to Word and PDF as single pages or complete case packages. Parambil is built for medical malpractice, birth injury, nursing home negligence, mass torts and product liability, motor vehicle accidents and maritime injury, and positions itself as a neutral third-party vendor serving plaintiff and defense firms alike. It supports HIPAA compliance and signs business associate agreements on request. Named customers include Kline & Specter, Jenner & Block, Cory Watson and Meyers & Flowers. The company is Parambil Corp of New York, founded in 2023 by Sara Dwyer, Liam Gordon and Ralph Horwitz.
Capability grades
All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.
AI Centrality
How much of the product is actually AI. Whether the machine learning is the mechanism the buyer is paying for or a feature layered onto conventional software, and whether the vendor is specific about which is which.
There is no content asset and no conventional product underneath. Parambil licenses nothing and publishes nothing of its own; the material it works on is the customer's medical records. Every deliverable it sells is a model output: the chronology, the de-duplicated Provider Guide, the billing analysis, the missing-document detection, the agents that investigate a case end to end and draft complaints and discovery, and the natural-language query layer over the record. Remove the models and what remains is an upload folder. The company's own framing matches, describing the platform as automating the ingestion, synthesis and verification of unstructured medical records, and its comparator is not another software product but manual review by paralegals, legal nurse consultants and retired physicians. That is the clearest form this axis takes: the AI is not accelerating an existing product, it is the whole of what is bought. Verified 2 September 2026.
Citation Accuracy and Hallucination Disclosure
Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.
Grounding is real, specific and structurally verifiable. Every chronology entry is hyperlinked to its source page, in-text citations link each summary detail to the original document for instant validation, and the platform flags documents referenced in the record but missing from the file, which is a form of telling the user where the evidence base is incomplete rather than filling the gap. An accuracy figure is published and it is unusual: a 44 per cent increase in correct information against traditional methods, alongside 95 per cent faster and more than five million records reviewed. What is not published is the method behind it. No test set, comparator definition, sample, evaluation protocol or date accompanies the 44 per cent claim, and nothing states who conducted the measurement or against what baseline, so a buyer cannot check the single number that would most justify the purchase. Nothing addresses hallucination in the drafting layer, where the Drafting Agent produces complaints and discovery requests rather than retrieving from the record. Verified 2 September 2026.
Autonomy and Oversight Model
What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.
This is a high-autonomy product with almost no published oversight model, and the gap between the two is the finding. The agents are described as executing multi-step workflows and investigating cases without continuous prompting and without waiting for prompts, re-evaluating conclusions as new records arrive, with a Case Agent working end to end and a Drafting Agent generating complaints, demands, interrogatories and discovery requests. Against that, the published verification affordances are real but passive: every assertion carries a hyperlink to its source page, and incomplete evidence is flagged. Nothing states where a human is required, what an agent completes before a person sees it, what happens when the record is ambiguous or the agent cannot support a conclusion, or what a firm should check before a generated complaint is filed. The website terms of use are the only agreement published and they address site access rather than the platform, so no review obligation exists in any document. For a product drafting pleadings from medical evidence unprompted, the oversight model is the thing a buyer most needs and is the thing least described. Verified 2 September 2026.
Operational and Outcome Evidence
Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.
The reference base is strong and specific. Fourteen customer logos appear on the home page, including Kline & Specter, Jenner & Block, Cory Watson, Meyers & Flowers and Kershaw Talley Barlow, which is a serious roster in mass tort and medical malpractice rather than a set of unknown names. Four testimonials carry a name, a title and a firm, among them Jay Edelson, founder of Edelson PC, and Benjamin Present of Kline & Specter, whose comment is substantive rather than promotional in that it claims the platform surfaces what a paralegal, legal nurse consultant and physician expert would miss. Four figures are published: more than five million medical records reviewed, over 20,000 hours saved, 95 per cent faster than traditional methods, and a 44 per cent increase in correct information. What holds this below the top band is that not one figure carries a method, a date or a named customer. No case study attaches an outcome to any of the fourteen firms, the hours-saved and accuracy figures are aggregate and unsourced, and nothing is dated, so a buyer can see who uses it and what is claimed but cannot connect the two. Verified 2 September 2026.
Privilege and Confidentiality Posture
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
One limb is answered well and the rest are not answered at all, on a product handling the most sensitive category of case material. The model supply chain is named openly, with Anthropic identified for the agent layer and Perplexity for live web search, which is more candour than most records in this corpus offer. HIPAA support and a business associate agreement on request are published, which is the right regime for medical records and is a real commitment. Beyond that the posture is thin. Nothing addresses whether customer content is used for training, in either direction. Nothing describes segregation between customers or matters; the only access statement is that granular controls govern user roles and system-wide data visibility, which is permissioning inside an account rather than a boundary between them. Privilege and work product are not mentioned anywhere, and the attorney work product doctrine is the doctrine that actually protects a litigation chronology. Retention is vague, and the website terms state that user content may be deleted at any time without prior notice with no backup obligation. One specific exposure is undescribed: a live web search feature on a platform holding protected health information, with nothing published about what leaves the system when it runs. Verified 2 September 2026.
UPL and Professional Responsibility Posture
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point.
Nothing published addresses the professional responsibility questions this product raises. Checked the home page, solutions page, practice areas page, security page, terms of use in full and privacy policy in full on 2 September 2026. There is no statement that output is not legal advice, nothing about who may use the platform or under what supervision, no reference to competence or verification duties, and no jurisdictional statement. The closest thing is a positioning claim rather than a professional one: Parambil describes itself as a neutral, third-party software vendor and an independent technology platform delivering fact-based insights to all parties, which addresses partisanship rather than the practice of law. The absence carries more weight here than on a research tool. The Drafting Agent generates complaints, demands, interrogatories and discovery requests, and the chronologies are described as courtroom-ready and as surfacing missed diagnoses and causation links, which is analysis close to expert opinion. A platform producing filings and causation analysis for plaintiff and defense firms, publishing nothing at all on the advice line, sits at the bottom of this axis. Verified 2 September 2026.
AI Governance and Bias Disclosure
Published governance over model behaviour: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
One sentence exists and nothing stands behind it. The security page states that Parambil leverages an advisory board of leading experts from large institutions and academia to help build trusted AI systems. That names a structure without naming a member, a remit, a cadence, an output or a single decision it governs. Nothing else was located: no responsible AI page, no governance framework or principles, no individual or function accountable for model behaviour, no account of what is tested before a model or agent change ships, and no certification such as ISO 42001. Nothing anywhere addresses uneven performance, which matters unusually here because the inputs are medical records and the outputs feed damages and causation analysis, so differences in transcription and extraction quality across handwriting, facility, record format or patient population would translate directly into case value. Checked the home page, solutions page, practice areas page, security page, terms of use and privacy policy on 2 September 2026. Verified 2 September 2026.
AI Safety and Data Stewardship
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
Two elements are published clearly and three are not. Access control and protection are stated: multi-factor authentication enforced across all user and admin accounts, granular permissions governing user roles and data visibility, and encryption of all sensitive data at rest and in transit. HIPAA support with a business associate agreement on request is published, which is the governing regime for this data class and is a genuine commitment. What is missing is the rest. No subprocessor is named anywhere: the security page refers to verified enterprise-grade cloud infrastructure partners and to regularly auditing vendors without identifying one, and the privacy policy lists categories such as hosting, cloud services and IT providers rather than entities. No incident or breach notification practice was located on any surface. Retention is vague in the privacy policy and adverse in the terms of use, which state that user content may be deleted at any time without prior notice and that the company has no obligation to back it up. The privacy policy itself, dated October 2023 and covering the Platform by its own terms, never once mentions medical records, protected health information or HIPAA, so the published document does not reach the data the product exists to process. Verified 2 September 2026.
AI Liability and Recourse
What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.
No agreement governing the platform is published, so nothing published answers who bears the loss when the output is wrong. The only liability text located is the website terms of use, version 1.0 dated 26 March 2024, and its scope is stated on its face: it governs use of the website, and section 2.1 licenses the site for the reader's own personal, noncommercial use, which is not a description of a platform bought by law firms. Under the rule that credit follows scope, that document cannot be read as the platform's allocation of risk. Recorded for completeness rather than credited: it caps the company's total liability at fifty US dollars for any cause whatsoever, disclaims all warranties including accuracy, gives no indemnity to the customer while requiring the user to indemnify the company, and compels individual JAMS arbitration with class and jury waivers subject to a thirty day opt-out. No master services agreement, subscription agreement or business associate agreement is published, and the BAA is available only on request. A firm handing over client medical records therefore cannot see, before contacting sales, any cap, indemnity, warranty or insurance position that would apply to it. Verified 2 September 2026.
Practice Systems Integration Depth
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
No integration into the systems legal work already lives in was located, and none is claimed. Checked the home page and its full feature list, the solutions page, practice areas page, security page, terms of use and privacy policy on 2 September 2026. No case management or document management system is named, no connector or integration page exists, and no API or developer documentation was located. What the platform offers instead is export: chronologies, notes, billing ledgers and data tables download to Word or PDF, as single pages or complete case packages. That is a file leaving the system rather than two systems staying in step. The gap is conspicuous in this lane, where the neighbouring records are the case management systems a plaintiff firm already runs, so a firm using Parambil moves records in and documents out by hand at both ends. Verified 2 September 2026.
Deployment Model and Data Residency
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
Cloud delivery is evident and neither limb this axis asks for is properly answered. The only residency statement located is in the privacy policy, which tells the reader that personal information will be transferred from their location to the company's facilities and servers in the United States. That is a country, stated once, in a document about personal information rather than about the medical records the platform processes, and it is the whole of what is published. No cloud provider is named, the security page referring only to verified enterprise-grade cloud infrastructure partners. No region or residency option is offered, nothing distinguishes processing location from storage location, and no tenancy or isolation model is described, so nothing states whether the platform is multi-tenant or how one firm's case files are separated from another's. For a product ingesting protected health information under business associate agreements, where data location is a compliance question rather than a preference, this is thinner than the product class warrants. Verified 2 September 2026.
Security Certifications and Trust Center
Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.
Two seals are displayed and neither is accompanied by a statement that would let a buyer test it. A SOC 2 badge and a HIPAA badge appear as images at the top of the security page. No SOC 2 type is stated, so it is not established whether this is Type I or Type II, and no auditor, certificate number, report date, coverage period, scope or trust services criteria appear anywhere. No route to obtain a report is offered, whether self-serve, on request or under non-disclosure, and there is no trust centre or portal. The HIPAA seal sits beside a carefully hedged sentence, that the company supports HIPAA compliance and signs business associate agreements upon client request, which is accurate framing since HIPAA is a regime rather than a certification, but it means the badge asserts more than the text does. No penetration testing or vulnerability assessment programme is described. What is published is a graphic; what this axis asks for is evidence. Verified 2 September 2026.
Model Supply Chain Disclosure
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
Two providers are named openly in product copy, which is more than most records in this corpus manage and is done without hedging. The agent layer is stated to work through a partnership with Anthropic, and the live web search through a partnership with Perplexity, each named at the point where the capability is described rather than buried in a policy. That tells a buyer whose technology handles the medical record and whose handles the outbound literature search, which is the more useful half of this axis. The rest is absent. No model or version is identified for either provider, nothing states where inference runs or under what terms, nothing addresses what either provider may retain, and no commitment exists to notify customers when the provider set or the models change. The infrastructure layer is unnamed entirely, described only as verified enterprise-grade cloud infrastructure partners. The Perplexity relationship carries a disclosure question the published material does not reach, since a live outbound search from a platform holding protected health information raises the question of what is contained in the query. Verified 2 September 2026.
Commercial Transparency
Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.
Nothing about cost is published. No pricing page exists, no tier or plan is named, no unit of charge is identified, and no structure is described, whether per user, per case, per page, per record or by subscription. The only commercial route on the site is Book a Demo, which appears in the header, the footer and at the end of every section. Nothing states whether implementation, onboarding or training is charged separately, and no trial or free tier is offered. A buyer cannot learn from anything published whether this is priced per matter or per seat, let alone what it costs. No pricing row is written, because the rule that triggers one requires published structure and none exists here. Checked the home page, solutions page, practice areas page, security page, about page, terms of use and privacy policy on 2 September 2026. Verified 2 September 2026.
Firm and Practice Coverage
Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.
Practice coverage is enumerated properly rather than gestured at. Six areas each have their own section with tailored treatment: medical malpractice, birth injury, nursing home negligence, mass torts and product liability, motor vehicle accidents, and admiralty and maritime injury, with the maritime template described as the personal injury framework adapted for Jones Act and vessel incident work. That is a real statement of where the product is built to operate. Both sides of the bar are named explicitly and repeatedly, with the platform described as trusted by plaintiff and defense firms and as an independent platform serving all parties, which is unusual in this lane and is borne out by a roster spanning mass tort plaintiff firms and a large defense firm. What is missing is the boundary. No firm size band is stated, nothing addresses solo or small firm suitability against mass tort portfolios running to thousands of claimants, no jurisdiction is named although United States practice is implied throughout, and nothing states what the product does not support. Verified 2 September 2026.
Legal Signals
What each signal meansA signal records what public sources say on the date shown. It is not a grade and it is not a recommendation. Where a signal reads Not addressed, it means the index did not locate the material in public sources on that date, which is a statement about disclosure rather than about the product.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
No located term or policy addresses the question either way.
Checked the home page, solutions page, practice areas page, security page, terms of use in full and privacy policy in full on 2 September 2026. Nothing addresses whether customer content is used to train or improve models, in either direction. No clause names training, model training, machine learning or model improvement. Two provisions bear on it without reaching it, and a reader should see both. The privacy policy permits use of personal information to provide, administer, maintain, improve and analyse the services and to develop new features and services, and separately permits aggregation of personal information to improve and add features, with aggregated information shared with third parties or published. The website terms of use grant the company an irrevocable, royalty-free, worldwide, sublicensable licence over user content to reproduce, prepare derivative works of, incorporate into other works and otherwise exploit it, expressly limited to the purpose of including that content in the Site. Neither names training, so under the naming test neither moves the value. The gap is sharper here than on most records because no platform agreement or business associate agreement is published at all, so the document where a training position would sit does not exist publicly.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Retention is acknowledged in public materials with no stated period.
Retention is acknowledged and no period is attached anywhere. The privacy policy states that personal information is kept for as long as necessary for the purposes it was collected for, unless a longer period is required or permitted by law, and adds that it may be retained for legitimate business interests including compliance, dispute resolution and record keeping, with deletion or anonymisation when no longer needed and isolation from further processing where deletion is impossible because of backup archives. No figure appears. The website terms point the other way and are quoted above: section 3.1 states that the company is not obligated to back up any user content and that content may be deleted at any time without prior notice, placing the burden of keeping copies on the user, and section 8 adds that terminating an account may involve deletion of associated user content from live databases. Nothing states how long chronologies, agent outputs, queries or uploaded records persist, and no retention setting is offered. Deletion is available on request by email. The absence matters more than usual because the material is medical records held under a business associate agreement that is not published.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
No located public material addresses walls or matter level segregation.
Checked the home page, solutions page, practice areas page, security page, terms of use and privacy policy on 2 September 2026. Nothing describes segregation between customers, between matters, or between users where a conflict requires it. The one access statement published is that granular access controls govern user roles and system-wide data visibility, which describes permissioning inside a customer account rather than a boundary between accounts, and system-wide data visibility as a phrase reads toward breadth rather than containment. Nothing states whether the platform is multi-tenant, how one firm's case files are isolated from another's, or whether a wall can be configured within a firm. The product does organise work into cases and folders, so the structure for separation exists, but nothing presents it as a security boundary. The gap is pointed in this practice area: mass tort work routinely involves firms on both sides of the same litigation, and Parambil markets to plaintiff and defense firms simultaneously as a neutral platform.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Published terms or policy address disclosure to authorities or in response to legal process, and no commitment or reservation regarding customer notice is located anywhere. The vendor has told the customer that data can leave and has said nothing about whether the customer hears of it.
The position is stated expressly and it is the adverse one. Section 3.1 of the privacy policy opens with the quoted words, qualified only by the phrase unless required by the law, and then lists the circumstances, which include disclosure where required by law or in the good faith belief that it is necessary to comply with a legal obligation including national security or law enforcement requirements, to protect the company's rights or property, to prevent fraud, to act in urgent circumstances, or to protect against legal liability. Several of those limbs are the company's own determination rather than a compulsion. No notice commitment appears anywhere, no undertaking to seek confidential treatment or to give the customer an opportunity to intervene, and no transparency report was located. The website terms of use add nothing on the point. This is one of the clearer statements of the position in the corpus, and it applies to a platform holding plaintiffs' medical records.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
No located public material identifies the corpus behind the product’s answers.
Checked the home page, solutions page, practice areas page, security page, terms of use and privacy policy on 2 September 2026. Parambil ships no corpus of its own and none is claimed. The material it works on is the customer's: medical records, billing records and case files uploaded by the firm, with the customer warranting its rights to that content under the website terms. The provenance risks this signal tracks therefore sit with the firm rather than the vendor. One dependency is worth recording because it is the exception. The live web search feature is described as giving real-time access to clinical guidelines and authoritative medical literature through a partnership with Perplexity, with every citation verified. That is third-party content entering the work product, and nothing published identifies which guideline bodies or literature sources are within scope, on what licence or terms that material is retrieved and reproduced, or what the verification of citations consists of.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
No located public material addresses whether authority is checked for subsequent history.
Checked the home page, solutions page, practice areas page and terms of use on 2 September 2026. No public material addresses subsequent history, treatment flags or good law checking, and none is claimed. The product does not retrieve primary law: it reads medical records and drafts from them, so a citator has nothing to operate on and the limb does not bite in its usual form. The analogous question does arise and is also unaddressed. The Drafting Agent generates complaints, demands, interrogatories and discovery requests, which are documents that cite law, and the live web search retrieves clinical guidelines, which are superseded and revised in much the way authority is. Nothing published states whether a retrieved guideline is checked for currency or whether a generated filing's legal citations are verified against anything.
Refusal and Uncertainty Behaviour
What does the product do when the answer is not in the corpus?
No located public material addresses what the product does when it cannot ground an answer.
Checked the home page, solutions page, practice areas page, security page and terms of use on 2 September 2026. Nothing describes what the system does when it cannot ground an answer, and no confidence or grounding indicator is described as shown to the user. Two adjacent behaviours are published and are recorded here because they are the closest the material comes without meeting the signal. The platform flags documents referenced within the record but missing from the file, and it visualises gaps in the medical record, which tells a user the evidence base is incomplete rather than telling them the model is uncertain. Company statements elsewhere describe claims being flagged when evidence is incomplete and conclusions re-evaluated as new records arrive. All of that concerns the completeness of the source material. Nothing addresses the agent that drafts a complaint or answers a query when the record does not support the conclusion sought.
Fabricated Citation Record
Does a public court record exist involving output from this product?
No court order, opinion or disciplinary record naming this product has been located as of the date shown. This is a statement about the public record, not a finding about the product.
Searched the AI Hallucination Cases database maintained by Damien Charlotin, and reporting drawing on it, on 2 September 2026 on the product and corporate name Parambil and Parambil Corp. No court order, opinion or disciplinary record naming the product was located. This is a statement about the public record rather than a finding about the product. One structural note bounds it in an unusual direction. The tracker records fabricated legal citations in filings, and Parambil's primary output is a factual chronology drawn from medical records rather than legal authority, so its characteristic failure mode would be a misstated clinical fact, a wrong date or a missed record rather than an invented case. That failure would surface as an evidentiary dispute or an expert challenge rather than as a sanctions order, and no tracker records it. The Drafting Agent, which produces filings, is the part of the product this signal would reach in its usual form.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
No located public material engages with bar or ethics guidance.
Checked the home page, solutions page, practice areas page, security page, terms of use and privacy policy on 2 September 2026. No public material engages guidance from any professional body governing the platform's users. ABA Formal Opinion 512 is not named, no state bar opinion is cited, and nothing addresses the competence, supervision or candour duties of a lawyer relying on a machine-generated chronology or a machine-drafted pleading. The company does engage an external regime, but a different one: it supports HIPAA compliance and signs business associate agreements, which governs the handling of protected health information rather than the professional obligations of the lawyers using the output. The one governance-adjacent structure mentioned, an advisory board of experts from institutions and academia, is described as helping build trusted AI systems rather than as engaging professional rules.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials claim time savings without addressing billing or disclosure.
Savings are quantified and the bill is not addressed. The home page publishes more than 20,000 hours saved and 95 per cent faster than traditional methods, and frames the commercial case explicitly around case economics, arguing that time is money and that improving the internal rate of return on cases will become a greater priority. Nothing addresses how AI-assisted work is recorded, billed or disclosed to a client, and no per matter record of AI-assisted work was located. The gap has a particular shape in this practice area. The buyers are contingency-fee plaintiff firms, so the hours saved are not billed to a client in the usual sense; the displaced cost is case expense, the paralegals, legal nurse consultants and retired physicians the platform is explicitly compared against, and that expense is customarily charged to the client's recovery. Nothing published addresses whether or how platform cost passes through as a case expense, which is the question a plaintiff's client would have.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
No located public material supports a client side disclosure obligation.
Checked the home page, solutions page, security page, terms of use in full and privacy policy in full on 2 September 2026. No subprocessor list exists in any form. The security page refers to verified enterprise-grade cloud infrastructure partners and to regularly auditing vendors without identifying a single entity, and the privacy policy lists categories only, covering hosting providers, cloud service providers, other information technology services providers, event management, email and newsletter services, advertising and marketing services and web analytics. Two entities are named elsewhere on the site, Anthropic and Perplexity, but they appear in product copy describing capabilities rather than in any list a firm could forward, and neither is presented as a subprocessor. No data processing agreement is published and no forwardable client-facing disclosure material was located. The one document a firm could obtain is a business associate agreement, available on request, which addresses protected health information rather than the disclosure a client's outside counsel guidelines would ask for.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
No located public material addresses court disclosure or verification certification.
Checked the home page and its full feature list, the solutions page, practice areas page, security page and terms of use on 2 September 2026. Nothing addresses judicial standing orders, disclosure of AI use, or certification that output was verified by a person. The platform produces artifacts close to what such a record would require and does not describe them as one: every chronology entry hyperlinks to its source page, missing and referenced-but-absent documents are flagged, and complete case packages export with notes, ledgers and data tables. Nothing captures which model or agent produced a given passage, when, or who reviewed it. The gap is sharpest at the drafting layer, since the Drafting Agent generates complaints and discovery requests that are filed with courts, and the chronologies are marketed as courtroom-ready, so the artifacts most likely to attract a disclosure question are the ones with no disclosure record attached.