A

Alexi

The AI is the product. The vendor describes itself as a private single tenant AI engine, built originally to generate evidence backed research memos, with a proprietary model layer and a retrieval first architecture. Remove the models and there is no product: memo generation, precedent finding, chronology building and workflow automation are all model driven.

Accuracy is independently measured with a published method. Alexi submitted to the 2025 Vals Legal AI Report legal research study, run by an outside evaluator that published its rubric and weights in advance. Accuracy carried 50 percent, authoritativeness 40 percent and appropriateness 10 percent, across 200 US legal research questions from attorneys at named firms. In the evaluator's published results, Alexi scored a 77 percent weighted aggregate, second of the four systems measured, against a lawyer baseline of 69 percent.

The architecture is described as retrieval first, to ground output in primary law rather than model memory. Two qualifications apply, though the figures themselves are independent and can be checked. First, the vendor's own account is selective. It leads with 80 percent overall accuracy against a 71 percent lawyer baseline, and 79 percent accuracy on single jurisdiction questions described as tied for the highest. It does not state that it placed second on the weighted aggregate.

Second, several major vendors declined to take part and at least one withheld permission to publish, so the field is self selected. All systems measured struggled on questions spanning several jurisdictions and trailed a generalist model on fifty state statutory surveys.

Autonomy is claimed and oversight is asserted without a described mechanism. The product runs workflows of several steps that carry legal principles and citations through successive stages. The vendor states that consistent, verifiable accuracy is what makes that automation safe, which is an argument for trusting the system rather than a control structure around it. The CEO has spoken publicly about human oversight in the context of a long term vision for AI assisted arbitration.

The vendor site, the FAQ, the workflow and product pages and the blog were searched. No description is published of what a workflow does unaided, a threshold at which it stops or escalates, a review surface given to a lawyer, or what the vendor commits to when an output is wrong. The product's stated advantage is that accuracy carries through each stage of an automated chain, so the absence of a published stopping rule is the main gap.

Customer claims stand in for evidence. The vendor states it is trusted by leading law firms and names buyer roles: CIOs and CTOs, knowledge management leaders, innovation teams, practice group leaders and managing partners. That is a market description rather than a deployment record. The vendor site, the blog, the FAQ and the news announcements were searched. No named customer is published, no case study pairing an organization with figures and a date, and no method that could be assessed.

The strongest quantified material is benchmark performance rather than production outcome. The Vals figures show how the system performs on a test set, not what changed at a firm that deployed it.

Two of the three parts are fully answered. On training, the vendor's FAQ states that nothing leaves the firm's private environment and that a firm's prompts, documents and workflow data never train global models and never mix with other firms. A document management partner's listing separately records no training on customer data. That answers both the training and pooling questions in plain terms. Segregation is the strongest element.

Every firm on the private cloud model receives a fully isolated single tenant instance, and the vendor lists what is isolated: retrieval, workflows, memory, security protocols and governance. Naming memory matters, because a platform that learns from firm usage could otherwise carry one firm's accumulated intelligence toward another. Supporting controls are AES-256 encryption, TLS 1.2 or higher, single sign on, role based access control and audit logs, under SOC 2 certification. The third part is not answered: no retention period or deletion commitment is published.

The audience is clearly lawyers and law firms, with named buyer roles inside firms and no consumer or non lawyer surface located. The vendor engages publicly with the problem of hallucination in filings and treats accuracy as a question of professional risk. The vendor site, the FAQ, the blog and the product pages were searched. No published position is published on the advice line, on competence or supervision duties, or on jurisdiction limits, although the product spans Canadian and US law, where the professional rules differ. The position can be inferred from who the product is sold to, but it is not published.

Principles and evaluation results are published without a governance mechanism. The vendor submitted to an independent third party evaluation and published the results, and it argues publicly that transparency and accountability in legal AI require exactly that rather than marketing claims. Submitting to outside measurement is a governance act, but it is an evaluation at one point in time rather than a management system.

The vendor site, the FAQ, the blog and the news announcements were searched. No AI governance framework, AI management certification such as ISO 42001, named owner of model governance or testing gate before release is published, and nothing addresses uneven output across matter types, parties or populations.

The vendor's security page and FAQ supply the control detail. Published: AES-256 encryption at rest, TLS 1.2 or higher in transit, single sign on, role based access control, audit logs, support for private VPC deployment, and deployments in specific regions for firms operating across offices. Architectural isolation is the strongest element: a fully isolated single tenant instance per firm covering retrieval, workflows, memory, security protocols and governance.

SOC 2 certification underpins the control set, and a Trust Center offers a request route to a security addendum, architecture whitepaper, data processing agreement and audit reports. Not located: a stated retention period or deletion control for research queries, uploaded case files and generated memos, a named subprocessor list, and an incident or breach notification practice.

No published indemnity, liability cap, carve out, warranty on output or insurance position is published, and no customer terms of service or master agreement is published as published on the property. Recorded as a pure absence on the surfaces reached. Worth noting the shape of the gap: this vendor stakes its positioning on measured accuracy and on reducing the risk of hallucinated citations reaching a filing, and publishes nothing about who bears the loss if one does.

One integration is named and none is documented. A product integration with iManage is announced, which is the document management connector that matters most for a research product used inside firms. The vendor site, the FAQ, the product pages and the news announcements were searched. No integrations index is published, no other named connector for practice management, court filing or Microsoft Word, and no documentation of what the iManage integration moves, in which direction, or what an administrator configures. The connection is announced rather than documented.

The deployment model is stated clearly, with partial residency detail. Two models are published. A private single tenant environment per firm is the default, and private cloud deployment is offered for firms wanting complete ownership and control of their own data. Single tenancy is stated as architecture rather than as an enterprise upsell. Geography is missing. The vendor site, the FAQ and the product pages were searched, and no named regions, customer selectable residency, or statement of where processing happens as distinct from where data is stored is published. That matters for a vendor operating across Canadian and US jurisdictions.

The vendor publishes a dedicated security page stating SOC 2 certification and operates a Trust Center reached through a request access flow. It states the Trust Center carries a security addendum, an architecture whitepaper, a data processing agreement and audit reports. That is a self serve request route to substantive diligence material rather than a sales gate. SOC 2 certification is also stated on the vendor's platform pages, about page and FAQ, and in a document management partner's technology listing.

Supporting controls published on the FAQ: AES-256 encryption, TLS 1.2 or higher, single sign on, role based access control, audit logs and support for private VPC deployment. No coverage period, audit scope, report date, SOC 2 type or named auditing firm is published. The certification is asserted on marketing pages, and no certificate was seen.

The vendor refers to proprietary models without identifying what sits underneath them. Published: a proprietary model layer the vendor brands as Alex, described as trained specifically on law rather than general purpose, and a retrieval first architecture. That tells a buyer the shape of the stack and that a specialist layer exists. Searched the vendor site, the FAQ, the product pages and the blog and located no named foundation model or provider underneath the proprietary layer, no statement of where models run, no subprocessor list, and no commitment to notify customers when the supply chain changes. A private single tenant environment narrows the exposure question without answering it.

No pricing page is published, no rate is published, no unit of charge is stated and no tier structure appears. Every commercial path located terminates in a demo or contact request. No free trial or self serve entry point is published, and no third party pricing figure is published either.

Segment and practice coverage are described with substance, without stated boundaries. The buyer is law firms, and the vendor names the roles inside them: CIOs and CTOs, knowledge management leaders, innovation teams, practice group leaders and managing partners. Practice coverage spans litigation, transactional, operational and administrative work, with a published treatment of personal injury covering case file summarization, chronology building and memo drafting.

Jurisdictional scope is Canada and the United States. Not located: any statement of which firm sizes or practice areas the product is not built for, and any list of coverage at state or provincial level, which for a research product is the boundary a buyer needs most.