Cooley teams up with Google to build, not buy, its document redaction AI: Big Law changes strategy

By Zendoric · October 9, 2026.

Cooley, one of the largest law firms in the United States, announced on Thursday that it is partnering with Google to co-develop an artificial intelligence agent to help redact and conceal confidential information in legal documents, according to Legaltech News (Law.com). The tool is built on Gemini Enterprise, Google's platform for deploying AI agents in corporate environments, and is, according to the source, one of several AI projects Cooley has under way in parallel.

The process being automated — known in the legal world as "redaction" — consists of locating and concealing privileged, confidential or sensitive information within a document before handing it over to an opposing party, a court or a regulator. In litigation of any size, it can mean reviewing tens of thousands of pages, a task that traditionally falls to junior associates, paralegals or lawyers hired specifically for document review in e-discovery proceedings.

According to Legaltech News, Cooley identified document redaction as a use case with "strong potential" for agentic AI, but one that required considerable engineering work; hence going directly to Google rather than to one of the legaltech vendors already offering automated redaction features. Broadly speaking, e-discovery platforms such as Relativity, Everlaw and DISCO have been incorporating this kind of capability for some time, which suggests Cooley was after something bespoke that those generic tools did not cover.

As industry context, the deal comes amid a wave of similar moves that Legaltech News itself reported that same week: ArentFox Schiff shut down its internal chatbot, AFSChat, acknowledging that third-party tools had already surpassed it in capability; Debevoise was rethinking how to bill for access to its proprietary AI platform, STAAR, beyond the billable hour; and Legora, one of the most visible legal AI vendors, launched "Skills" features to customize agents with each firm's internal knowledge. The picture these pieces paint, read together, is of a sector that is no longer debating whether to adopt AI, but under what model to do so: build bespoke with a hyperscaler, buy from a legaltech specialist, charge for what you build yourself, or simply backtrack when the in-house tool stops being competitive.

Our reading is that this kind of deal marks a new phase in the relationship between Big Law and Big Tech. Over the past two years, large firms have mostly limited themselves to licensing third-party products — Harvey, Legora, CoCounsel — built on models from OpenAI, Anthropic or Google. That a firm the size of Cooley is sitting down to co-design the product directly with Google, rather than merely buying it, suggests some firms are beginning to see bespoke AI as a source of competitive advantage over their peers, not just an internal efficiency gain. The trade-off is tighter dependence on a specific infrastructure provider: if Gemini Enterprise does not evolve at the pace the firm needs, Cooley will have invested its own engineering in an asset it does not fully control.

This connects to something we have been pointing out about AI's impact on legal employment: the firm pyramid is narrowing at the base. Document redaction is, by definition, routine, high-volume work — exactly the kind of task that falls first to automation. In the short term, that puts pressure on entry-level document review roles, where generations of junior associates have learned the trade by poring over case files. In the long term, if the underlying thesis holds, that same automation frees up hours for what truly sets a firm apart — legal judgment, litigation strategy, the client relationship — and brings down the cost of legal services currently out of reach for individuals and small businesses. The transition, meanwhile, will not be painless for those building their careers today reviewing documents.