G

Genie AI

The models drive the core capability, but they sit on a document product that would function without them. Drafting, playbook review, redlining, comparison across documents and tabular extraction are all generative, and the Eidetic Intelligence architecture is what the vendor sells against general purpose LLMs. Underneath sits a conventional platform: a docx editor, PDF import, document sharing with privacy controls, department spaces and a library of more than 500 templates.

All of these appear on the Free plan, and Genie shipped the templates as an open source library in December 2018, before the assistant existed. Remove the models and a template repository with an editor remains.

Accuracy is measured and published, but as comparative scores rather than citation accuracy. The benchmarking program carries two dated head to head reports: 18 February 2026 against Claude Sonnet on a regulatory scenario across several jurisdictions, and 22 April 2026 against Claude CoWork on a commercial supply agreement. They score 15 and 10 dimensions respectively, with rationale per metric that the vendor says a reader can reproduce or critique.

Those are comparative scores on Genie's own rubric, scored by Genie, rather than a measured citation accuracy figure or a hallucination rate with the test set described. Three headline marketing numbers match neither report and are not published on the benchmarks index. They are 92 percent accuracy rated by lawyers, a 140 percent advantage over GPT and Claude, and 90 percent on a 65 document study against 79.3 percent for Claude CoWork and 37.3 percent for ChatGPT.

Genie's home page states that the architecture prevents hallucinations before they affect the final result, an unmeasured claim. The benchmarks page also carries the meta description 'Independent head-to-head benchmarks' while its body says Genie's own engineering team publishes them.

Autonomy is claimed prominently and the oversight mechanism is not described. Marketing sells agents that draft, review and negotiate contracts autonomously, invites the buyer to control when they need to and delegate when they do not, and describes agents handling complex deals across several documents and borders. What is published about the review point sits in the agreement rather than the product: clause 3.4 disclaims any guarantee that the software comprehends what is put into it, clause 5.9 states that recommendations, suggestions or actions taken by the software are not legally binding, clause 6.10 prohibits use for automated decision making with legal or similarly significant effects, and clause 21.7 instructs the user to have a human review each output as appropriate.

Those are allocations of responsibility, not a described control structure, and no threshold, review surface or escalation path is published on the home page, the use case pages or the model training page. The direction of the gap is worth noting: the marketing sells delegation while the agreement prohibits the automated decision making that delegation implies.

Named customers with named individuals and at least one figure are published, without dates or method. The customer stories index carries attributed video case studies. Jude Legg, Compliance Lead at Firefish, speaks to cutting contract review time by more than 50 percent; Jafeth Bulsink, Head of Sales at Bright Renewables, to 25 salespeople negotiating contracts with no legal department; Neil Rowe, Head of Commercial at Cambridge United, to player signings; and Nazmus Tareque of AnthroTek to closing a seed round without a lawyer.

A logo strip of 13 named organizations, including hyperexponential, Blockworks and the Jamaica Stock Exchange, sits alongside them, and the model training page states that Clifford Chance, Pinsent Masons and Withers ran pilots from 2018. No deployment date, scope or measurement method is published.

The published commitment and the published agreement point in opposite directions on training, and the agreement is what a customer is held to. The security center states that customer documents, prompts and confidential data are never used to train shared or internal AI models, and the model training page is headlined on never training on your contracts. Clause 7.3 of the product terms grants Genie a nonexclusive, irrevocable, worldwide, perpetual license to use, copy, display, distribute and reproduce a customer's Documents to improve the software and, in terms, its algorithms and models.

Clause 6.8 lets Genie use Your Data and Documents to make Improvements, and clause 7.5 vests ownership of those Improvements in Genie. The model training page's FAQ narrows its promise to contracts never training models that other customers see. On the other side, OpenAI and Anthropic are named and operate under zero data retention agreements, a DPA and a subprocessor list are published, and clause 20.7 says communications are not covered by legal professional privilege. No tenant or matter level segregation model and no retention period is published.

A specific published position on the advice line exists, without supervision duties or other jurisdictions. Clause 20 of the product terms states that Genie AI Limited is not a law firm and is not authorized or regulated by any body regulating legal services. It cannot provide reserved legal services, including litigation, probate, notarization and administering oaths, and operates only in general commercial matters.

It then sets out the consequences for the buyer: no requirement to hold professional indemnity insurance, no right to complain to a legal ombudsman, no access to a compensation fund such as the SRA's, and no legal professional privilege over what the user receives. Clause 6.7 excludes domestic, personal and noncommercial use, so there is no consumer product requiring disclosure. Nothing published addresses a lawyer's own competence and supervision duties.

The whole treatment is drawn from the England and Wales regulatory perimeter, on a product sold across more than 150 jurisdictions including the United States. No equivalent statement for any of them is published.

A described quality mechanism exists where a governance position should be. The model training page sets out a five layer account in which qualified lawyers contribute training data, set prompt engineering standards, supervise retrieval logic and rate every release against a panel benchmark before it ships. The security center publishes a specific EU AI Act position: Genie is not a high risk system under the Annex III categories, the obligations applying to it came into force on 2 August 2026, and it maintains AI literacy measures for staff operating the system.

No individual or role inside the vendor is named as accountable for model behavior, no prerelease testing regime is published beyond accuracy benchmarking, and nothing is published on uneven output across matter types, counterparties or populations. The ISO 27001 certification covers information security, a different subject.

Substantive published policy covers almost all of this ground, missing a retention period. The security center publishes a register of 35 controls grouped as an auditor would walk them. They include AES-256 at rest and TLS 1.2 or higher in transit, quarterly access reviews recorded in an Access Rights Review Log, a joiner, mover and leaver process, environment segregation, peer reviewed code with deployment separated from development, and tested encrypted backups.

A business continuity plan carries a recovery time objective of one hour for critical services, and incident response is documented, with incidents tracked through an Incident and Corrective Action Log. Breach practice is stated: notification to affected customers without undue delay and, where Genie is controller, a report to the UK Information Commissioner's Office within 72 hours. A current subprocessor list names 15 third parties with the purpose, processing location and security measures for each.

The register states only that formal procedures govern the retention and secure disposal of customer data, and no retention period for documents or prompts is published.

A full allocation of loss is published, with nothing covering the output itself. Clause 18.2 caps aggregate liability at the fees paid or payable in the 12 months before the first event giving rise to the claim, and clause 2.4 sets a separate cap of one thousand pounds for the Free Package. Clause 9 gives the customer a defense and indemnity against third party claims that use of the software infringes intellectual property rights.

Conditions sit in 9.2, a remedy sequence in 9.3 of procuring a license, modifying the software or terminating, and six named exceptions in 9.4, including Your Data and any use under a Free Package. There is no recourse when the output is wrong. Clause 5.8 excludes liability for reliance on the software's outputs, recommendations or actions, clause 21.3 disclaims any warranty that content is accurate or complete, and clause 21.4 excludes all liability for reliance.

Clause 20.4 states that Genie is not required to hold professional indemnity insurance, and it makes no claim to hold any.

Integration categories are named, and no documentation an implementer could use is published. Genie's home page lists MCP connectors, an API, and native integrations with email, CRM, cloud storage and electronic signature, and the Enterprise plan adds SSO alongside API and MCP integration. A Word add in has its own page. Every 'Learn more' link on the home page integration card resolves to an anchor rather than a destination, and no developer index, integration directory or help center documentation is published from the site navigation or the footer.

No document management integration such as iManage or NetDocuments, matter management, electronic billing or court filing connection is named. That fits a product sold to commercial teams rather than firms, but leaves the depth unverifiable.

Residency is published in detail, and the tenancy model is not stated. The security center states that the platform runs on Amazon Web Services with customer documents stored in UK and EU data centers. The subprocessor table gives a processing location for each of the 15 third parties, which separates where data sits from where it is processed: AWS and MongoDB in the UK and EU, CloudAMQP in the EU, and OpenAI and Anthropic in US and EU data centers.

Nothing published states whether the platform is multi tenant or single tenant, and no private or on premises option and no region choice for the customer is published. The same subprocessor table lists Google Cloud and Stripe at global data centers, which is broader than the UK and EU statement on the security page.

Certification is real, current and documented in detail, with the report itself held back. The security center names the standard as ISO/IEC 27001:2022, gives certificate number 310012019 and names the issuing body as Citation ISO Certification. It publishes the original approval date of 7 February 2019, a current certificate dated 7 February 2026, expiry on 6 February 2029, a last external audit on 29 April 2026 and a last management review in June 2026, with the page marked last reviewed 30 July 2026.

The vendor states plainly that it does not hold a SOC 2 report. The certificate, the ISMS scope and the Statement of Applicability are available on request under NDA through a named security mailbox with a stated response within two working days. They are not offered as a download, so the ISMS scope is withheld. The control register refers to independent penetration testing after significant changes but names no testing partner and publishes no summary.

The providers are named, and the change commitment runs the wrong way. OpenAI and Anthropic are both listed on the published subprocessor table with their purpose, processing locations in US and EU data centers and stated security measures. The security center adds that both operate under Zero Data Retention agreements, so inputs and outputs are not retained by the provider after processing. The model training page states that Genie passes legal context at inference time instead of fine tuning, specifically so newer foundation models can be swapped in as they land.

No specific model or version is identified for either provider. On notification, the subprocessor page says changes will be posted there and asks readers to check back regularly, which leaves noticing a supply chain change to the customer.

Real pricing is published across the whole range, including a starting price for the enterprise plan. The pricing block on the home page carries a figure, a unit and a term for all three plans. Free is $0 for one user with limited tokens and no document exports, Pro is from $75 per month for one user with enough usage for roughly five documents, and Enterprise is from $600 per month for unlimited users and unlimited tokens.

Each plan's contents are itemized, and signup needs no sales call and no credit card. Every figure carries a 'From', so each is a starting price rather than a rate. Nothing published states what onboarding, the dedicated customer success manager or the bespoke AI training program bundled into Enterprise add. Prices display in US dollars while clause 13.3 of the product terms states that payment will be made in pounds sterling unless stated otherwise.

Coverage is documented precisely on the dimensions the vendor sells, and the limits are stated in the agreement. Dedicated pages exist for six user groups including in house lawyers, procurement, sales, finance and founders, for startup, mid market and enterprise companies, and for six industries: energy, construction, technology, real estate, mining and sport. The Enterprise plan states coverage of more than 150 jurisdictions and ten languages.

Clause 20.2 of the product terms names what Genie cannot do, reserved legal services including litigation, probate, notarization and administering oaths, and confines the product to general commercial matters. Clause 6.7 excludes domestic, personal and noncommercial use. No law firm segmentation is published, nothing addresses government or court use, and the practice dimension is expressed as industry verticals rather than areas of law, so the sectors served are known but not which law is covered.