S

SimpleLegal

The models drive one core capability, on a platform that would function without them. InvoiceAI is a named, shipped product: machine learning that reviews invoice lines alongside existing rules engines and keeps learning from corrections. Electronic billing, matter intake and management, vendor management, budgets, accruals, reporting and the CounselGO portal all stand without any model, and the platform sold on exactly that from 2013 until the AI layer arrived in 2021.

The owner's current marketing calls the product AI native, which does not fit a platform that ran for eight years before its AI capability was introduced.

The method is described in detail, and no measurement is published. The owner says InvoiceAI analyzes historical and live invoices with machine learning trained on millions of legal invoice charges, and feeds its findings into existing rules engines so that learned and fixed rules work together. The issue types are listed, from block billing and vague descriptions to work done by the wrong staff class. Every flag attaches to a specific invoice line the reviewer can open against their billing guidelines, so verification is built into the work.

No accuracy figure, false positive rate, test set, evaluation method or independent benchmark is published. The published figures measure savings found, which says nothing about what was missed or wrongly flagged.

Human review is built into the learning loop, with no published thresholds. The owner describes InvoiceAI as handling the first pass review of incoming bills within a framework that learns as reviewers correct invoices, so the person sits at the correction point and the corrections become the training signal. Approval workflows sit between a flag and any payment decision. Not located: any threshold at which a flag applies automatically rather than going to review, whether a reduction can take effect without human approval, and what the vendor commits to when a flag is wrong, which matters because a wrong flag cuts a law firm's payment.

Quantified figures are published without named customers. The owner states more than 550 corporate legal departments and $5.2bn in annual legal spend processed as at May 2026, a measure of work flowing through the system rather than seats sold. Outcome figures are unattributed: a customer quote cites 10 percent savings identified in legal spend, and the owner reported an average of six figures in travel billing identified across customers.

Independent reviews appear on several platforms. No named customer paired with figures and a date, and no case study with an assessable method, is published.

Nothing published says how client confidences are handled, and the product's design makes the question pressing. There was no statement on training with customer content, no retention or deletion terms, no segregation model, and no treatment of privilege or work product. The owner's own description has the models learning from customer corrections, so training on customer data is the stated design, and nothing published says whose data it is, whether it is pooled across customers, or whether a customer can decline. Invoice narratives describe what lawyers did on matters, material that can be privileged.

The audience is corporate, and no position on advice is published. Users are in house legal departments, legal operations and finance teams, with outside counsel working through the CounselGO portal, and the product analyzes billing rather than giving legal advice. No published position on advice versus tooling, treatment of competence or supervision duties, or jurisdiction limits is published.

No AI governance position is published: no framework, principles document, management certification, named owner of model governance, testing regime before release or bias disclosure. The gap has an edge here. InvoiceAI judges whether billed work was appropriate, including whether it was done by the correct staff class, and it learns from historical invoice data. Nothing published addresses whether its judgments differ across firms, practice areas, staffing models or timekeeper seniority.

A security posture is published at owner level, and nothing confirms what applies to this product. Onit runs a trust center on the Vanta platform, reachable without a sales conversation. Its control detail renders client side and was not retrieved, so no specific control is confirmed for SimpleLegal. The product pages, the owner's material and press coverage, searched, gave no retention period, deletion control, encryption statement, hosting disclosure, named subprocessor list or incident practice.

Because the invoice review models keep learning from customer corrections, what is retained and for how long is a live question the located material does not answer.

Who bears the loss from a wrong flag is not addressed. No indemnity, liability cap, carve out, warranty on output or insurance position is published, and no customer terms of service were reached on the surfaces searched. The structure of the product sharpens the gap. The AI's output reduces payments to a law firm that is not the vendor's customer and has no contract with it, so the party that bears the direct cost of a wrong output has no route to recourse.

Integration is claimed by category rather than by connector. The vendor states that integration with enterprise resource planning and finance systems gives live visibility into spend and performance, and third party material describes support for multiple systems. CounselGO is a real named component, though it is a portal the product ships rather than a link into another system. No integrations page, named connector, API documentation or statement of supported billing format standards is published.

Nothing on the deployment model is published: no hosting provider, named regions, residency choice for customers, tenancy model, or statement of where processing happens.

A trust center exists, and no certification is confirmed for this product. Onit operates it at trust.onit.com on the Vanta platform, with a controls section and a stated commitment to protecting customer data, and it is reachable without a sales conversation. The owner also publishes SOC 2 Type 2 attestation for a sibling product, a virtual data room, along with AWS hosting, role based permissions, two factor authentication and encryption.

The trust center renders its content client side, and its certification list was not retrieved. No coverage period, audit scope, report date or auditing firm is published for SimpleLegal, and the sibling's SOC 2 was not read across.

The vendor's own model layer is described, and nothing beneath it is identified. InvoiceAI is presented as proprietary machine learning developed by the owner rather than licensed in, and the owner has built or acquired several other AI products. No foundation model or provider, location where models run, subprocessor list or commitment to notify customers of supply chain changes is published. The detailed AI description dates from 2021, and current product pages describe AI capability without naming InvoiceAI, so what powers the product today is not confirmed.

No pricing is published. No pricing page, rate, unit of charge or tier structure is published, and third party sources confirm the absence, one stating there was no public pricing as at May 2026 and another that plans are customized and quoted. Every commercial path ends in a demo or quote request, on a platform sold on visibility and control over legal spend.

The segment is described with substance, and the owner publishes the boundary itself. The buyer is in house legal teams that need matter management, electronic billing and spend visibility without enterprise implementation complexity. The owner positions this product against its own enterprise platform, so a prospect is told which of the two fits. Practice scope is consistent: spend, matters, vendors and reporting, with no claim to advisory or drafting capability. Not located: an industry list, organization size bands in numbers, jurisdictional coverage, or language support.