Google Cloud has launched Gemini Enterprise for Legal with Freshfields, Cleary and Weil

Google Cloud has launched Gemini Enterprise for Legal, its first dedicated legal AI offering for the broader market, with Cleary, Freshfields, Weil, and Williams & Connolly among the launch customers.

It arrives in a category where the specialists got there first and have the revenue to show for it, with Legora reaching $100mn in 18 months and European challengers raising to build the layer underneath.

The product is organised around four pieces rather than a single assistant. There are purpose-built legal skills, connectors into the systems firms already run, a partner ecosystem, and a governance layer with audit logging and centralised IT controls.

Those skills cover the tasks that consume junior time. Brief drafting, citation verification, contract lifecycle management, regulatory scanning, and fulfilment of data subject access requests are all named, alongside document redaction and converting legacy agreements into reusable playbooks.

The connector list is the more revealing part. Gemini Enterprise for Legal integrates with iManage, NetDocuments, Thomson Reuters, RelativityOne, and Harvey, which means Google is wiring itself into products that compete with it as readily as into the document management systems that do not.

Legora is going further and integrating directly, so that a response can link back to the relevant Legora thread with citations, source grounding, and permissions intact. That is a notable arrangement for a company Google might otherwise be trying to displace.

It also says something about how this market is settling. Law firms have already bought specialist tools and are not going to rip them out, so the realistic play for a hyperscaler is to sit above the stack rather than replace it.

DSAR fulfilment is a quietly significant inclusion for European readers. Responding to data subject access requests under GDPR is expensive, deadline-bound, and largely mechanical, which makes it one of the clearest cases where automation saves real money rather than demonstrating a capability.

Accenture, Deloitte, and KPMG are in the partner ecosystem, which is how enterprise software of this kind actually reaches large firms.

Nobody at a magic circle practice deploys an agentic platform without a systems integrator involved, and the consultancies bring the change management that the technology alone does not.

Google has not disclosed pricing. That omission is standard for enterprise launches and also unhelpful, since cost per seat is the number that determines whether a mid-sized firm can follow the launch customers.

The governance framework is doing more work than it might appear. Audit logging and centralised controls address the objection that has slowed legal AI adoption more than accuracy has, which is that a firm cannot deploy a tool it cannot supervise when client confidentiality and privilege are at stake.

Citation verification appearing as a named skill is its own quiet admission. The legal profession’s experience of generative AI has been shaped by fabricated case law reaching actual filings, and a product that checks citations is a product acknowledging the problem exists.

The launch firms are worth reading as a signal too. Cleary, Freshfields, Weil, and Williams & Connolly are not experimental buyers, and firms of that standing agreeing to be named at launch is a marketing asset Google presumably had to work for.

The competitive picture is now genuinely crowded. Anthropic has put Claude into Microsoft Word with contract review leading its use cases, OpenAI is selling into the same accounts, and European specialists such as Lexroom in civil-law jurisdictions are building for legal systems the US-centric products handle poorly.

None of which guarantees adoption. Legal is a conservative buyer, partner compensation models do not reward efficiency in obvious ways, and a tool that makes associates faster has an awkward relationship with the billable hour.

What Google brings that the specialists cannot is the rest of the stack. A firm already running Workspace and Google Cloud has a procurement path that a standalone vendor has to argue for, and that advantage tends to matter more in enterprise software than product quality does.