How It Works Research Exclusions Matter Memory Heppner Alternatives Pricing Join Waitlist
Marcella. Legal Intelligence. Verified.
Independent Market Analysis · August 2026

Legal AI Positioning Map

A structural read of the legal AI market following ILTACON 2026, and an honest placement of Marcella within it. Two axes, because the conventional one does not describe the contest that matters.

Prepared by Technology Outlaws LLC Period May to Aug 2026 Sources Public announcements & trade press Status Published analysis
About This Analysis

Who made this, and how

This is a self-produced report. We build Marcella, and Marcella is one of the vendors plotted here. Read it with that in mind. Every placement is drawn from publicly announced capability between May and August 2026, and every source is a vendor's own published material or trade press coverage of it. No vendor was consulted and no vendor paid for or influenced any placement.

The format is familiar on purpose. It is not a product of any analyst firm, and it reproduces none of their research or methodology.

Two panels rather than one, because a single chart cannot carry the argument. The first uses the conventional framing a buyer already carries into the room, breadth of capability against scale in market, and places our own product where that framing honestly puts it, which is low. The second re-plots the market on the two properties that determine whether an output survives judicial scrutiny. A self-assessment section further on scores our own product against fourteen criteria, including the ones we fail.

Anyone can check the vendor claims here against the announcements they came from. Every externally sourced claim below is cryptographically sealed, and you can verify the chain without trusting us.

How this document is sealed

This document names competitors, so every externally sourced claim in it is cryptographically sealed. The receipt chain records each claim with the source it was read from and the date it was read. Any alteration to a claim, its source, or their order breaks the chain and the verifier says where.

It was also drafted with an AI system. That is precisely why the seals exist. The tool that wrote it belongs to the same class of tool this document argues cannot be relied on without a verification layer underneath, so we put one underneath. The seal is produced by the same attestation architecture that runs inside Marcella, and the ceiling is the same in both places: it attests that the record is unaltered. It does not attest that any claim is true. Open the sources and check them.

Chain parameters
FieldValue
Formatdraft-farley-acta-signed-receipts
Structurecompound (nested sub-attestations under one outer frame per section)
SignatureEd25519 over RFC 8785 (JCS) canonicalized payload
ChainSHA-256 prev-receipt hash, genesis 64 zeros
Issuermcf:issuer:8M9L7TqHEMjR
Public key6d2ad7c652b9829d9900f3dc4b9cf4e87027b50535627e8f02869c4de15755a4
Sealed claims22, across 3 compound frames
Chain head4840f0ce16c71040d015aa45605ec0ae10d5b040f8dc603b717d0b0763b66dbc

Download the receipt chain   Download the verifier

The verifier needs only Python and the cryptography package. No account, no network, and no Technology Outlaws software. Run python3 verify-provenance.py provenance.json.

What carries no seal, and why. Every judgment in this document carries no seal. The quadrant placements and the fourteen component scores are the author's assessment. An assessment has no source URL, so it cannot be sealed, and it is not. Remaining rows of the feature matrix are sealed as each source is confirmed. Where you see no seal, you are reading an opinion or a claim whose source has not yet been verified, and the document tells you which.

What a seal cannot do. It does not prevent a wrong claim. A fabricated statement with a plausible-looking URL would seal perfectly if nobody opened the URL. What the chain enforces is that no claim ships without a source attached, and that nothing changed after sealing. One assertion in an earlier draft of this document described AT&T as a Fortune 10 company. It is not. That sentence had no source that supported it, so it could not be sealed, so it did not ship. That is the mechanism working, and it is the entire argument of this document applied to itself.

Panel 1

The conventional view

Stated capability against market scale, the shape a procurement committee will recognize. On these axes Marcella is early and narrow, and it is not close. Anyone who draws a chart like this with themselves in the top right corner is selling something. Note the marker: every other vendor belongs to one category, and Marcella is the only entrant plotted in all three at once. The scorecard further down publishes the component scores behind the placement, including the ones that do not flatter.

Panel 1 · Vision × Execution
Post-ILTACON 2026. Placement reflects publicly announced capability and market presence.
Scaled, narrow Scaled, broad Early, narrow Early, broad Breadth of stated capability Scale in market Thomson Reuters LexisNexis Clio + vLex Harvey Legora Filevine Anthropic Eve Litera iManage NetDocuments Relativity DISCO Marcella
Content owners / incumbent platforms AI-native entrants DMS, drafting & discovery incumbents Marcella: all three categories, plus a security plane no swatch exists for

Our own placement is low, and it belongs there. These axes measure revenue, installed base, support scale, territory and channel maturity. Against companies with thousands of employees and decades of installed base, a young product scores low on all of it. That is the correct result and we are not going to argue with it.

What the axes do not measure is whether the output holds up when someone checks it. That is the subject of the next panel.

Panel 2

The axes that decide a sanctions motion

Verification depth against custody and provenance control. These are the two properties that determine whether an output holds up when a court, a regulator, a disciplinary body or an insurer examines it. The market has not organized itself around them, which is why the corner is occupied by one entrant and no one else.

Panel 2 · Verification × Custody
Horizontal: where verification happens in the pipeline. Vertical: who controls the data and can attest to the stack.
Governed, unverified Evidentiary grade Assertive Verified, uncustodied Post-hoc lookup to pre-generation substrate Custody & Provenance iManage NetDocuments Litera Thomson Reuters LexisNexis Anthropic Clio + vLex Legora Filevine Harvey Eve Relativity / DISCO Marcella
Content owners / incumbent platforms AI-native entrants DMS, drafting & discovery incumbents Marcella: all three categories, plus a security plane no swatch exists for

The verification and custody stack is built. Components are marked Live where they are in service, Built where the code is complete but the capability has not been switched on in production, and Untested where the artifact is produced but has not yet been adjudicated.

Two qualifications are stated deliberately rather than buried. The self-authenticating exhibit credential is generated today; no court has yet ruled on one. Nothing has entered the record. And the device-forensics tier is complete but not yet running against production GPU. Both distinctions are load-bearing. A claim of judicial acceptance that cannot survive a general counsel's first question destroys the credibility of every other row on this page.

Capability Matrix

Feature breakdown

Capabilities as publicly announced between May and August 2026, stated as of 28 August 2026. Where a vendor announced a capability but has not shipped it generally, the cell reflects the announcement. This market moves quickly and some of these cells will be out of date within a quarter.

Vendor Headline move Orchestration
layer
Own /
derived model
Persistent
matter memory
Drafting
layer
Citation
verification
Customer-tenant
deployment
Exit &
portability
Audit of already-
filed documents
Court-verifiable
credential
Infra / security
layer
Thomson Reuters CoCounsel Legal GA; "Thomson" in-house LLM YesYes, open-weight baseMatter-centricBrief Builder Post-hoc, KeyCite-groundedNoVendor-heldNoNoNo
LexisNexis Legal Intelligence Engine harness Yes, dynamicOwn models + frontierContext carry-forwardAgentic drafting Post-hoc, Shepard's-groundedNoVendor-heldNoNoNo
Harvey Harvey II; Tenet model; AIUC-1 Yes, agenticYes, Kimi K3 baseYes, core featureYes Model-sideNoVendor-heldNoNoEthical walls via partner
Clio + vLex Clio Work; bar member program Skills infrastructureFrontier modelsPractice dataClio Draft Corpus-groundedNoVendor-heldNoNoNo
Filevine LOIS console; LOIS Explore free tier Yes, acts on mattersFrontier modelsSystem of recordLOIS for Word Verified citations claimedNoVendor-heldDocument analysis onlyNoNo
Legora Legora aOS agentic operating system YesFrontier modelsWorkflow contextEditor / Word Model-sideNoVendor-heldNoNoNo
Eve EveOS; Atlas data layer; Analyst YesFrontier modelsAtlas, portableDemands, discovery Not a focusNoStructured, exportableNoNoNo
Litera Firm AI Search via Lito; Foundation 365 Single agentFrontier modelsFirm experience dataDraft / Lito Deterministic heritageNoVendor-heldNoNoNo
iManage Next-gen platform GA; MCP read/write Via MCPNoKnowledge + contextWord editing agent NoGoverned, hostedExport toolingNoNoGovernance controls
NetDocuments Legal context graph; MCP Via MCPNoContext graphWord editing agent NoGoverned, hostedExport toolingNoNoGovernance controls
Anthropic Claude for Legal: plugins & connectors Yes, agenticFrontier ownerSession / projectClaude for Word No native citatorAPI / enterpriseNo retention by defaultNoNoNo
Relativity / DISCO Agentic review; AI bundled free Yes, discovery scopeFrontier modelsMatter workspaceNo Out of scopeNoVendor-heldReview, not integrityNoNo
Marcella Retrieval-first + attestation substrate MoA orchestration Live Frontier via own tenant Matter memory Live SCDL: conditioning, no training Live Pre-generation anchoring Live Runs in customer tenant Live Never transferred; firm holds keys Live Forensic audit of filings Live Produced today No ruling yet Endpoint, entropy, hardware Live
Compare & Contrast

Where the architectures actually diverge

The prevailing architecture

Shared by every vendor on the chart above.

  • Generate first, then check. A citation is produced by a model and validated afterwards by lookup against an index.
  • Verification is probabilistic where it is thorough. A second model invocation judges whether the source supports the proposition, reintroducing the error class the check exists to catch.
  • The link between output and the retrieval context is discarded at generation time, so no system can answer whether an authority was actually before it when it drafted the passage.
  • Attestation is decoupled from verification. A flagged response can be cleaned and then attested, laundering the event out of the record.
  • No measured error rate exists, because outcomes are not recorded as counts. There is no defensible per-tenant fabrication figure.
  • Multi-tenant SaaS custody. The vendor holds the data; the firm holds a contract.
  • Exit is a feature that has to be built, because custody was transferred in the first place. The strongest position on offer is a vendor that structures the firm's data well enough to be exported and used elsewhere. That is a real improvement, and still an answer to a problem the architecture created.
  • Verification is a product feature, so it is only available to paying customers, which means no court can ever require it.
  • Drafting is trained or generic. Firm-specific voice is achieved by fine-tuning on the firm's own documents, which reads across every conflict screen, encodes privilege into weights that cannot be audited, and cannot forget a departed attorney. The alternative on offer is a generic drafter that produces prose in nobody's style.
  • Every drafting layer generates only into its own product. None will audit a competitor's output.

The Marcella architecture

Design intent. Components marked Spec are not yet shipped.

  • The verification substrate is created before generation, not inferred after it. A citation that cannot be anchored cannot be emitted.
  • Integrity is deterministic and model-free, using a Merkle root over the canonical corpus, where a single-byte change cascades. Drift is adjudicated separately from nonexistence.
  • Render policy is fixed at ingest from the source licence class, never decided at query time, so licensed text cannot leak through a per-query bug.
  • The human attestation act is a hard gate. Work product that is not attorney-verified cannot receive an attestation, with no partial and no override.
  • Telemetry records counts only, never citation strings or excerpts, producing a measured per-tenant unanchored-citation rate as an empirical error figure.
  • Deployment into the firm's own tenant. The data does not move; the compute comes to it.
  • Verification of a credential is free and open. Anyone can check one offline, at any time, without an account and without contacting us.
  • Drafting without training. Firm-shaped drafts are produced by conditioning a generic model on convergence-gated style derivatives (argument skeleton, clause-position map, citation-authority graph, redline posture) abstracted across a minimum threshold of same-class documents, with the source discarded after extraction. No model is trained on privileged data, so there is no ethical-wall breach, no privilege in weights, and no deletion problem when an attorney departs.
  • Attach mode. The same layer will audit the output of any third-party drafter against the firm's own frames and judicial-treatment data, with zero integration into that competitor's generation process. A firm can run a rival platform and still verify it here.
  • Attestation on the standard a court already applies. The verification checklist an attorney completes before work product can leave the system is built around the reliability factors courts use to evaluate expert and technical evidence under Daubert: whether the method can be tested, whether it has a known error rate, whether standards govern its operation, and whether it is generally accepted. The system produces a measured error rate precisely because that is one of the factors. This is not a compliance checkbox designed in the abstract. It is built to the test that will actually be applied if the output is ever challenged.
  • Attestation-gated export. Output is labelled machine-generated in a non-suppressible state and cannot leave the system until a human completes a full multi-item attestation, recorded as an immutable event. Partial attestation is rejected. Display is permitted before attestation; export is not.
  • Drafts are emitted as tracked changes rather than prose, so every machine edit is individually reviewable, acceptable and rejectable.
  • There is no exit to build, because there was no transfer. The data sits in the firm's own tenant under the firm's own keys for the entire relationship. Disconnection is revoking access to software, not repatriating an asset. A firm that wants its complete footprint before disconnecting takes a static export of material it already holds.
  • The sealed credential outlives the vendor. An exported credential is verifiable by anyone, offline, indefinitely, with no callback to the issuer and no dependency on the issuer continuing to exist. Portable data still requires someone to interpret the exporting vendor's structures; a self-authenticating artifact does not.
  • Deploys where the work already happens. Native to the firm's existing productivity suite rather than another destination to log into. A full private-tenant deployment on the customer's own infrastructure completes within a business day.
  • An infrastructure and security layer beneath the application that no content or application vendor on this chart operates.
Panel 3

The layer nobody else operates

Every vendor on this map is a content company or an application company. None of them runs an endpoint agent, an entropy plane, or a hardware attestation floor, because none of them is an infrastructure and security operator. This is the comparison that does not appear in any analyst chart, because no analyst covering legal software thinks to draw it.

What legal AI vendors secure

  • Application-layer access control, SSO and role permissions.
  • Encryption in transit and at rest, on infrastructure they rent.
  • Ethical walls and conflict screening, generally inherited from an integration partner rather than enforced natively.
  • Tenant separation inside a shared multi-tenant estate.
  • Certification against an AI assurance standard, which is a governance artifact rather than a control plane.
  • Data residency commitments expressed contractually.

All of it sits above the application. None of it survives a compromise below the application.

What Marcella secures

  • Heuristics plane. Adversarial static analysis as a pre-commit CI gate that blocks deploy on critical findings with no override. Content fingerprinting of every ingested corpus document against a native known-bad set, quarantining poisoned corpus before it is ever indexed. Runtime recon detection that removes the system from an attacker's view rather than returning an error.
  • Entropy plane. Concentric decoy broadcast across network, session and seed-derivation layers, so an attacker who pierces one ring finds an unbounded field of valid-looking targets at the next. Multi-stream attestation deriving a single seed without which no inference, unlock or signing operation proceeds. A hardware random floor beneath all of it with no software attack path.
  • Detection plane. Composite-score quarantine, plus threat intelligence derived from pattern absence and displacement rather than known signatures.
  • Endpoint plane. A managed device agent and a behavioral entropy gate where the biometric baseline never leaves the device. Only a hashed entropy contribution does.
  • Deployment. Runs inside the customer's own tenant under managed identity with private endpoints and per-tenant key isolation. Derivative-only persistence is enforced on every write, so privileged source content is structurally incapable of entering the record.
  • Supply-chain hygiene. Copyleft dependency audit and personally-identifiable-data scanning are hard gates before any binary is packaged. No artifact ships without both passing.
Provenance

The question no vendor on this map can answer

In a single quarter, two of the largest vendors in legal moved from renting frontier models to post-training their own from open-weight foundations. The economics are compelling and the engineering is legitimate. The procurement consequence has not been priced in.

Fact
Documented position
Consequence for a regulated buyer
Open-weight is not open-source
Open-weight releases publish trained parameters only. Training code, training data and methodology are not published.
A derived model cannot be audited for what went into it. The buyer inherits an unexaminable dependency regardless of who published the weights or where.
Derived-base adoption is now mainstream
Two major legal AI vendors announced in-house models post-trained from open-weight foundations within the same week in August 2026.
This is a market-wide condition, not a single vendor's misstep. Any procurement questionnaire that asks only "which model do you use" is now asking the wrong question.
Licence terms create a live counterparty dependency
The Kimi K3 Licence permits derivative models but provides that if the licensee or an affiliate operates a Model as a Service business, and aggregate licensee-and-affiliate revenue exceeds USD 20 million over any consecutive twelve months, the licensee must enter a separate agreement with Moonshot AI before any commercial use. Whether a vertical application vendor is a MaaS operator under that definition is an open question, and the licence waives the requirement entirely for purely internal use. A separate clause, with no MaaS condition attached, requires any product above USD 20 million in monthly revenue to display “Kimi K3” prominently in its user interface.Sealed
The foundation beneath privileged work product is governed by a commercial agreement with a third party the buyer has no relationship with and cannot inspect. This is a contract exposure with a date on it, not a matter of sentiment.
Enterprise procurement is already screening
AT&T vice-president Mark Austin has stated the company is not currently using any Chinese open-weight models while it evaluates the associated risks, naming DeepSeek and Moonshot.Sealed
The screening exists today at the largest buyers. Government, defense-adjacent and CUI-handling procurement will apply it more strictly, not less.
No vendor publishes a model bill of materials
No platform on this map publishes an auditable inventory of every model, weight set, derived base, inference host and third-party dependency in its stack, with attestation.
This is the actual gap. It is nationality-neutral, it survives the next licence change and the next base model, and it is answerable only by an operator who controls the stack rather than rents it.

The durable framing is provenance, not geography. An argument built on where a model was trained ages the moment an equivalent base is published elsewhere, and reads as prejudice to a sophisticated buyer. An argument built on whether a vendor can produce a documented, auditable, attested inventory of its own stack does not age, applies uniformly to every vendor including this one, and is the exact question a court, an auditor, a disciplinary body and an insurer will each arrive at independently.

State the facts. Let the buyer reach the conclusion. That is the register this entire document is written in, and it is the only register that survives contact with opposing counsel.

Self-Assessment

The component scores, published

Most vendor comparisons show you a position without showing you the arithmetic. This one shows the arithmetic. Below is our own product scored against fourteen criteria written for this document, with the reasoning for each, including the criteria we fail. If you are evaluating us, these are the questions worth asking, and the answers are here before you have to ask them.

Sub-criterion
Score
Basis
What it can deliver today
Working capability in production
Strong
Verification substrate, drafting layer, security planes, protocol surface and customer-tenant deployment are all in service. Capability is not the constraint.
Product velocity
Strong
Multiple production layers and a substantial patent filing program delivered inside a single year. Shipping rate is materially above what headcount would predict.
Response time to a market change
Strong
No committee, no roadmap cycle, no board approval between a market signal and a shipped response.
Financial durability
Weak
Pre-revenue. Self-funded. No external capital and no runway modelled against a customer-acquisition curve.
Deployment velocity
Strong
Native to the productivity suite the firm already runs. A full deployment, including a private-tenant configuration on the customer's own infrastructure, completes inside a single business day. The process is a candidate for full automation, which would reduce onboarding cost per account close to zero.
Procurement path
Strong
Transaction through a hyperscaler commercial marketplace lets an enterprise buyer draw against an existing committed cloud spend agreement rather than opening a new vendor line. This routes around the slowest step in enterprise legal procurement and is unavailable to vendors selling direct.
Route to a signed contract
At launch
Pricing is published and the product is available now. What has not yet accumulated is a body of closed enterprise accounts, because the commercial phase has just opened. The interval between a first conversation and a running deployment is a day, not a quarter.
Reference customers and support record
Weak
No reference customers, no renewal history, no independently verifiable deployment at another organization. The single single most valuable gap on this table: one reference account converts the weakest column into the strongest available proof.
Support capacity
Strong
Support is live around the clock, with a four-hour response commitment and direct escalation to the principal engineer at any hour. Tailored support arrangements are available for firms with specific coverage or response requirements. Firms accustomed to enterprise vendors will recognize the difference between a four-hour commitment honored by the person who wrote the system and a four-hour commitment honored by a queue.
Operating capacity
Strong
We run our own business on the architecture we sell. Every function, from engineering to patent prosecution to go-to-market, is executed by specialized agent teams built on the same orchestration layer offered to clients, each grounded in curated expert sources rather than general training data, and each able to scale on demand without a hiring cycle. That is not a claim about what the technology could do for a firm one day. It is how this company operates today, and it is the reason a small organization delivers at the rate this document describes. We are not asking any firm to hand its practice to autonomous agents. We are demonstrating what the underlying architecture supports by running on it ourselves first.
Partner network & channel
Building
Approved independent software vendor, partner, reseller and developer within a major cloud partner program, with global distribution onboarding underway. The partner program is in active development. The economics are unusual in this category: a customer with committed cloud spend can apply it to this purchase, which means the sale lands against a commitment the customer has already made rather than a new budget line.
Where it is going
Technical novelty
Strong
The only architecture on this map placing verification before generation, and the only one producing a third-party-verifiable evidentiary artifact.
Roadmap coherence
Strong
Composes with rival platforms at both ends of their pipeline rather than requiring displacement.
Read on the buyer
Strong
Two decades of enterprise technology sales with direct focus on legal operations, state and local government, and large enterprise accounts, including capital and operating expense structuring on the buyer's side of the table. The regulatory direction was also read correctly and early. The gap is transactional validation, not comprehension of the market.
Revenue model
Strong
Verification is free and open to anyone. Generation is paid, and pricing is published rather than quoted. A firm can see what it costs before speaking to anyone.
Go-to-market plan
Strong
Building natively inside the productivity suite the firm already runs means the product reaches customers through an existing global sales and partner organization rather than through a channel that has to be constructed first. Distribution is inherited, not built.
Jurisdictional coverage
Strong
Corpus coverage extends beyond domestic sources to international legal and official gazette material. The conventional geographic-reach criterion measures sales offices, which is a poor proxy in this category: what a firm needs is authority coverage in the jurisdictions its matters touch, and that is a corpus question rather than a real-estate question. Depth in a jurisdiction is the product.
Net position
Visionary
Highest innovation score on the map against the smallest operating base on the map. The placement sits where it does because capability, deployment velocity and procurement path are genuine execution strengths, while support depth and proven customer outcomes are genuine execution gaps. It does not cross the midline, and will not, until an installed base and a public reference list exist. That is the whole gate, and neither is an architecture problem. Neither is claimed here before it is true.

The honest one-line read: a finished product at the beginning of its commercial life. The capability is built and running. What has not accumulated yet is the reference list, and a buyer is entitled to weigh that. What a buyer gets in exchange for going early is a roadmap still responsive to them, published pricing, and a verification standard that remains checkable by anyone whether or not this company is in the room.

See it against your own filings. A short paid assessment runs a brief your firm has already filed through the verification layer and returns what it finds. No deployment, no connector, no IT ticket. You supply one document and read the report.

What that buyer gets in exchange is a capability no funded competitor offers, at a stage where the roadmap is still responsive to them, and with an open verification standard that does not strand them if the company does not survive. That last point is not a consolation. A verifiable credential that anyone can check offline, forever, without calling back to the vendor, is the only architecture on this map that degrades gracefully if its author disappears.

Panel 4

Both ends of someone else's pipeline

Every other vendor on this map competes for the same seat: replace what the firm uses today. One position on the board is structurally different: a layer a firm can adopt without displacing anything, which attaches to the front of a rival's pipeline, the back of it, or both.

Upstream: the retrieval and context source

Their agent calls in.

  • An open protocol tool surface exposing the same substrate three ways: direct point reads, pre-assembled compound context in a single call, and graph-aware semantic traversal across typed relationships.
  • One provenance record per compound invocation, enumerating every node assembled into the response. That is chain of custody over an assembled subgraph rather than element by element.
  • Context assembled once and served warm to a second verifying pass, collapsing the cost of a two-model proposer-verifier loop that is otherwise prohibitive at production scale.
  • Any protocol-capable client can consume it. Two document management incumbents shipped protocol surfaces in August 2026 with no verification layer behind them.

Downstream: the audit gate

Their output comes back.

  • Attach mode ingests the finished output of any third-party drafting or research system and evaluates it against the firm's own derivative style frames and judicial-treatment data.
  • Zero integration into the third party's generation process. No API agreement, no partnership, no permission. The firm supplies the document.
  • Produces a structured delta report, which terminates at the same attestation gate as native output. It stays machine-generated until a human clears the full checklist, and no export before that.
  • Extends to documents already filed, where no prospective control was ever applied.

A firm does not have to choose. This layer runs alongside whatever research or drafting platform is already in place, and it can also run alone. Document management and knowledge platforms are the natural fit, since they hold the firm's content and have recently exposed the protocol surfaces this connects to.

The firm holds the controls, not a vendor. The verification client runs on the firm's side, so no software provider can decide whether the firm's own work gets checked. That is the same rule that governs the attestation layer, for the same reason.

Structural Findings

What the map says

Orchestration is table stakes
Every vendor mapped announced an orchestration or operating layer within one quarter. A capability held by twelve competitors is a qualification requirement, not a differentiator, and should not lead any positioning document.
Memory is now contested, not owned
Persistent matter context shipped across four independent vendors in a single quarter, so a firm evaluating platforms should expect it as standard rather than treat it as a differentiator. The question worth asking instead is where that accumulated context physically resides, who can reach it, and what happens to it if the relationship ends.
Model sovereignty has arrived, cheaply
Two major vendors now run models post-trained from open-weight foundations rather than renting frontier APIs. This collapses the assumption that owning a corpus implies exclusive ability to act on it, and introduces an unauditable dependency into the foundation of privileged work. Regulated and government buyers are already screening for this, and which no vendor on this map can currently answer with documentation.
Price has no floor
Free tiers, bundled AI at no additional charge, and a state bar offering a platform as a complimentary member benefit all landed in the same period. Any positioning that depends on undercutting is already obsolete.
Verification is asserted, not architected
Every vendor claims verified citations. None has published an architecture in which verification precedes generation, is model-free, is gated on human attestation against the reliability factors a court actually applies, and produces a measured error rate. This is the only column in the capability matrix where the market is uniformly empty.
Two different portability promises, widely confused
A vendor made data portability a headline this quarter, structuring case data so it can be used outside the platform rather than locked inside it. That is a real advance and firms should give it credit. It is also worth understanding what it does and does not solve. What you gain: if you leave, your data comes with you in a usable form. What comes with it: the data still lived in the vendor's environment for the life of the relationship, which means it was subject to their access controls, their breach exposure, their retention practices and their subpoena surface the entire time. Portability is a good answer to the question of how you leave. It is not an answer to the question of who held your clients' privileged material while you stayed. The alternative is an architecture where the data never left the firm's own environment, which makes exit a matter of switching off software rather than recovering an asset. Both approaches are defensible. They protect against different things, and a firm should know which risk it is buying down.
Time-to-value is the unclaimed axis
Consider what adopting an enterprise legal platform actually looks like from inside a firm. Vendor evaluation runs a month or two. Security review and IT approval add several more, and at a firm with any regulated client work, that is a committee, a questionnaire, and a queue. Then connectors, data migration, taxonomy work, change management and training. Six months is a good outcome. Nine is common.

Now consider what is happening during those months. Attorneys at the firm are already using AI. They were using it before the evaluation started and they will keep using it while the committee meets, because the work is due regardless. Every one of those months is a month of unverified citations entering draft work product with no attestation, no record of what was checked, and no way to reconstruct afterward what the model actually had in front of it. The exposure does not pause for procurement. It compounds.

That is a present problem, and the conventional purchase does not begin to address it for two or three quarters. A deployment that completes inside a business day, inside the software the firm already runs, is not a marginally faster version of the same purchase. It is the difference between having a plan and having a fix. No announcement this quarter contested that ground.
Every vendor is competing for the same seat
Twelve platforms are asking a firm to replace something. Each announcement this quarter framed itself as the place the work now happens: the operating layer, the system of record, the one intuitive experience. That framing creates a displacement decision, a procurement cycle and an incumbent to fight. A layer that composes rather than displaces faces none of those, and can be adopted alongside a rival platform the firm has already paid for.
Drafting is universal; auditable drafting is not
Every vendor mapped ships a drafting capability, so drafting itself differentiates nothing. Two properties do. First, whether firm-specific output required training a model on privileged documents, which every firm-voice approach on the market does, at the cost of conflict-screen integrity and the ability to delete. Second, whether the drafting layer will evaluate output it did not generate. No vendor on this map will audit a competitor's draft, because each one's drafting layer exists to keep work inside its own product.
Nobody is defending below the application
Twelve vendors, zero endpoint agents, zero entropy planes, zero hardware attestation floors. Legal AI is staffed by content, product and application people. The security posture on offer is governance documentation and inherited platform controls, both of which describe a compromise rather than prevent one. This is not an oversight the market will correct quickly, because correcting it requires a different kind of company.
Regulation is moving toward the empty corner
Pending state legislation would convert professional-conduct guidance on AI verification into statutory duty enforced through court sanction and bar discipline. That is a mandate for exactly the property no vendor on this chart has built, and it arrives on a timeline none of them control.
Platform Scope

What else is in the box

The comparison above deliberately holds the ground where the market is empty. It understates the platform, because the verification layer sits on top of an infrastructure stack that a firm acquires in the same deployment. A summary, not a catalog.

Supply chain defense
The largest software compromises of recent years have not come through the front door. They arrived inside trusted dependencies, build pipelines and package registries, and they reached organizations far better resourced than any law firm. The stack includes adversarial static analysis as a blocking gate before code ships, content fingerprinting of ingested material against known-bad sets, and runtime detection that withdraws the system from an attacker's view rather than returning an informative error. A firm deploying the platform inherits this. A firm deploying the full stack inherits all of it.
Model traffic containment
A trained model layer that keeps a firm's language model traffic out of every provider's systems except the one the firm controls, with a training path that draws on frontier model capability without exposing the firm's material to it. The provenance question raised earlier in this document applies to us as well, and this is the answer to it.
Network and endpoint
Private network transport, managed device agents, hardware-anchored entropy, and behavioral gating in which the biometric baseline never leaves the device. Attestation is expressed against recognized federal control frameworks rather than a proprietary scheme.
Adversarial matter intelligence
Matter memory is usually described as a convenience feature. It is also a competitive asset. The same accumulated context that speeds a firm's own work supports analysis of how an opposing party, an opposing firm, or a particular authority has behaved across matters. That is a use case no vendor on this map has framed, because framing it requires the memory to be the firm's own rather than the vendor's.
Practice packs and billing
Domain packs for specific practice areas rather than a single general legal model, and native support for standard legal billing formats so that time and expense output moves into existing systems without a translation step.
Adjacent professional domains
The same verification substrate operates outside litigation support: tax and digital asset reporting, forensic integrity work applicable to financial crimes and fraud analysis, and corporate compliance and emissions reporting. Firms with transactional, regulatory or forensic practices are buying one platform rather than three.
Built on itself
Engineering, patent prosecution, marketing and go-to-market at this company run on the same orchestration architecture offered to clients. That is stated as evidence rather than as a selling point: the deployment speed, response times and delivery rate described in this document are produced by the system being described.
Next Step

Bring one filed brief

The fastest way to evaluate any claim in this document is to test it against work your firm has already produced. Nothing is deployed, no connector is installed, and no IT approval is required. You supply a document and read what comes back.

Book an evaluation

A working session against your own filings, with the verification layer, the drafting layer and the attestation gate running live. We will also walk the security stack for your IT and risk people, who will have questions this document does not answer.

marcellalegal.com/evaluation/

What the session covers

Your filed brief run through the verification layer, with every citation anchored or flagged. The drafting layer conditioned on your firm's own documents. The attestation gate, and the record it produces. Then the security and deployment questions your IT and risk people will raise, answered directly rather than deferred to a follow-up.

Pricing is published. You will not be asked to request a quote.