L
Luminary
Luminary is an AI platform for trust and estate planning that turns trusts, wills, amendments and entity documents into structured, source-cited records, then uses them for estate diagrams, beneficiary and tax modelling, planning insights and firm-branded client presentations. It is sold to wealth managers, estate planning law firms, trust administrators and tax advisors, with a dedicated offering for estate planning attorneys; it does not draft documents or give legal advice.
Luminary AI extracts and summarises document data with citations back to the source language, and an assistant answers questions about a household's plan; the company names OpenAI as an AI vendor under a data processing agreement and says neither it nor its AI vendors train on customer data. Luminary Platforms, Inc. is headquartered in New York and raised a $22 million Series A in September 2026. Estate planning has no lane of its own in this index; general-legal-assistants is the nearest.
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.
The models are the engine of the core capability, layered on a modelling and presentation platform that works without them, which is the B band. Luminary AI extracts and summarises trusts, wills, amendments and entity documents into a structured record of the household, and Insights and the assistant work from that record; the estate diagrams, deterministic tax engine, scenario modelling and firm-branded presentations run on the structured data whether it arrived by AI extraction or by manual entry, and the Subscription Terms offer household builds as a professional service. Verified 18 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 and documented with links to the source, short of accuracy figures an outsider can test, which is the B band. The AI Document Intelligence and Law firms pages say extracted data and assistant answers are source-cited, that every important detail can be traced back to the governing language in the document, and that customisable review controls let attorneys validate findings. A landing page claims over 98 per cent accuracy in data extraction but gives no test set or method.
The citator and primary-authority limbs do not bite on a product that reads the customer's own documents (R15). Verified 18 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.
A written commitment that professionals review AI output, with real review surfaces, short of thresholds or a named constraint, which is the B band. The vendor describes a human-in-the-loop interface that keeps the advisor in control, source citations and customisable review controls, and Insights that teams review, prioritise and track before anything reaches a client; the Subscription Terms make the advisor responsible for using independent judgement and involving legal professionals.
R124(2) was applied: no mode or tier carries a categorical limit, and no confidence threshold is published. Verified 18 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.
A named customer with figures but no method or date, which is the B band. The Caprock case study (a multi-family office with $13.8 billion under advisement and 29 advisors) reports more than five hours a week saved, a 25 per cent increase in advisor production capacity and onboarding more than a week faster, with the capacity figure given as the chief operating officer's estimate. Other named case studies (SCS, New Republic Partners, Bragg) carry quotes. Verified 18 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.
Substantive published commitments on confidentiality and training, short of privilege handling, which is the B band. The Subscription Terms make client information confidential, commit to notice before compelled disclosure, and require deletion of customer data 20 days after the term; the security and FAQ pages say neither Luminary nor its AI vendors train on customer data, that the AI retains nothing after processing, that a data processing agreement governs OpenAI, and that role-based permissions control collaboration with outside advisors. Privilege and work product in client documents are not addressed. Verified 18 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.
A plain position on advice versus tooling, short of the lawyer's competence and supervision duties, which is the B band. The Subscription Terms say the Services do not offer legal, tax, accounting or investment advice, that the advisor is responsible for involving legal professionals, and that Luminary is not a law firm and no attorney-client relationship arises even where attorneys work on its strategy support services; the Law firms FAQ says it does not draft documents or provide legal services.
The licence limits use to the United States. Nothing addresses attorneys' professional duties when relying on the output beyond verifying citations. Verified 18 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.
No AI governance position was located, which is the D band. Searched the Subscription Terms, AI Document Intelligence, Law firms, pricing, security and integrations pages on 18 September 2026. Nothing describes how Luminary AI is tested before release, who is accountable for it, or how extraction performs across document types or drafting styles; the security page's review of vendors' SOC 2 reports is a security control, not AI governance. Verified 18 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.
Published policy covering most of the ground, short of an incident practice and a full subprocessor list, which is the B band. The Subscription Terms let the advisor export and delete data and require Luminary to destroy all copies 20 days after the term; the security page describes AES-256 encryption at rest and in transit, role-based permissions, AWS storage in at least two US availability zones with regular backups and no offshore storage, and vendor due diligence; the FAQs say the AI retains no data after processing.
AWS and OpenAI are named as vendors, but no full subprocessor list or breach notification commitment was located. Verified 18 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.
A real published liability position with scope and caps, short of standing behind output accuracy, which is the B band. The Subscription Terms warrant that the platform will operate materially as documented, give an intellectual-property indemnity with defined scope, procedure and remedies, and cap liability at 12 months' fees with carve-outs for IP infringement, fraud and wilful misconduct. They disclaim data accuracy, make the advisor solely responsible for advice, and limit Luminary's liability for its AI and hosting vendors to its own gross negligence or wilful misconduct. Verified 18 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.
Real, documented integrations, but into wealth platforms rather than the systems legal work lives in, which holds the row at the B band. The integrations page says client profiles and entities sync with Addepar, Black Diamond and Orion portfolios, eMoney households import with financial data, Snowflake shares household, account and asset valuations nightly, and a public API exposes documents, entities, individuals and households; Salesforce is also named. No document management, practice management or other law-firm system is integrated. Verified 18 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.
Storage residency is stated without the processing location for the AI, which is the B band. The security page says data is stored on AWS in at least two US-based availability zones per region and that no data is stored offshore, and the licence limits use to the United States. Where OpenAI processes documents, and the tenancy model, are not stated (R38). Verified 18 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.
A named attestation with the report available on request, short of readable scope or dates, which is the B band. The security page says Luminary is SOC 2 Type 2 certified, undergoes independent audits and penetration testing, and offers its annual SOC 2 Type II report through a Drata trust centre on request. The trust centre returned only a loading page (R128), so no audit period or auditor was read. Verified 18 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.
The provider is named with data terms, short of the models or any change notice, which is the B band. The pricing FAQ names OpenAI as an AI vendor under a data processing agreement and says neither Luminary nor its AI vendors train on customer data, the Subscription Terms name OpenAI and AWS as vendors, and the Law firms FAQ refers to model provider partners in the plural. No model or version is named, the assistant's Fast and Thinking modes are not mapped to models, and no notice of changes is committed. Verified 18 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.
Pricing is by custom proposal while a tier name is published, which is the C band. The pricing page says the team prepares a customised proposal because pricing depends on factors specific to each team; a May 2025 release introduces Luminary Go as an accessible product tier, and the Subscription Terms tie licensed volume to authorised users and subscription tier and define a Luminary Client as a household. No figure or stated unit of charge is published. Verified 18 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.
Segments and scope are described with substance, short of boundaries on what the AI handles, which is the B band. Luminary publishes pages for wealth managers, estate planning law firms, trust administrators and tax advisors, says clients typically have taxable estates from $5 million to over $1 billion, lists the documents it processes (trusts, wills, business agreements, Crummey letters), and limits use to the United States; it states that it does not draft documents. Which document types or drafting conventions the AI handles poorly is not stated. Verified 18 September 2026.
4 public documents
The public pages on file for Luminary, with the recorded signals each one supports and the date it was last read. Open any of them and check the reading against the record.
-
withluminary.com/security2 signals
Client Data in Training, Ethical Walls and Matter Segregation
Read Sep 18, 2026
-
withluminary.com/subscription-terms2 signals
Prompt and Output Retention, Third Party Request and Subpoena Notice
Read Sep 18, 2026
-
withluminary.com/users/law-firms2 signals
Bar Guidance Alignment, Billing and Fee Posture
Read Sep 18, 2026
-
withluminary.com/pricing1 signal
Outside Counsel Guideline Readiness
Read Sep 18, 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?
A public policy or trust page states no training on customer content, with no matching term located in the published agreement.
Public pages state no training on customer content, with no matching term in the agreement. The security page, pricing FAQ and Law firms FAQ say neither Luminary nor its AI vendors train models on customer data, that the AI retains no data after processing, and that a data processing agreement governs OpenAI. The Subscription Terms license customer information to provide and improve the Services and let Luminary use de-identified Aggregate Data to improve its products, without naming training either way.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
The customer controls the retention window, by product configuration or by contractual instruction, but zero retention is not stated as available.
The customer controls deletion, with a contractual end point. The Subscription Terms make the advisor responsible for changes and deletions of its data, let it export at any time, and require Luminary to delete on instruction or destroy all copies 20 days after the term ends; the Law firms FAQ says the AI retains no data after workloads are processed. Zero retention is not offered as a platform setting.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Segregation is asserted in public materials with no published detail on how it is enforced.
Role-based permissions are asserted without published detail. The security page and FAQs say permissioned access lets trust officers, administrators, attorneys, tax professionals and wealth managers work on the same household data, but how permissions are set per household or enforced is not documented.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Terms commit to notice where lawfully permitted. No transparency report located.
Notice is committed in the contract. Section 7.2 of the Subscription Terms requires prompt written notice of any legal request or subpoena for the other party's confidential information, which includes client information, so the customer can seek a protective order, and limits any compelled disclosure to what counsel advises is required. No transparency report was located.
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.
Searched the AI Document Intelligence, Law firms and security pages and the Subscription Terms on 18 September 2026. The AI reads the customer's own estate documents, so no legal corpus is expected (R15); the tax law and rates behind the deterministic tax engine are not identified.
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.
Searched the same surfaces on 18 September 2026. Answers cite the customer's own documents rather than legal authority, so no check of subsequent history arises (R15).
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.
Searched the AI Document Intelligence and Law firms pages, the pricing and security pages and the Subscription Terms on 18 September 2026. Answers are source-cited and the assistant offers Fast and Thinking modes, but nothing describes what it does when the documents do not support an answer, and no confidence signal is described.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
No court order, opinion or disciplinary record addressing fabricated or hallucinated legal citations produced by this product has been located as of the date shown. This is a statement about the public record on that one subject, not a finding about the product, and this signal is not a litigation history.
Searched web and trade press reporting on court sanctions for AI-fabricated citations on 18 September 2026 for any court record addressing fabricated or hallucinated content in output from Luminary. None located. The product analyses estate documents and does not draft filings. This signal does not record litigation history of any other kind.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Public materials refer to professional responsibility in general terms without naming guidance.
Professional responsibility is referred to in general terms. The Law firms FAQ says Luminary does not provide legal services and that attorneys can verify AI output through source citations, and the Subscription Terms say it is not a law firm and creates no attorney-client relationship. No ethics opinion, court AI order or bar guidance is named.
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, and the product sits inside a fee relationship between a lawyer and a client where those savings would change the bill.
Time savings are claimed without addressing billing or disclosure. The Law firms page tells estate planning attorneys the AI moves them from document collection to assessment in a fraction of the time, and the Caprock case study reports five or more hours a week saved; nothing addresses how law firms should bill or disclose AI-assisted review to clients.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A current subprocessor or model provider list is published.
The AI provider is named, short of a disclosure pack. The pricing FAQ names OpenAI as an AI vendor under a data processing agreement, the Subscription Terms name OpenAI and AWS as vendors, and the security page says vendors are held to agreements. The Law firms FAQ refers to model provider partners in the plural, so the list may be partial, and no DPA or forwardable disclosure material is published. The Drata trust centre could not be read (R128).
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.
Searched the AI Document Intelligence and Law firms pages and the Subscription Terms on 18 September 2026. The outputs are planning analyses and client presentations rather than court filings, and nothing addresses disclosure of AI use to a court.