Google demonstrated its legal AI capabilities at a 'Google Legal' event on September 16, showing lawyers how Gemini Enterprise can handle tasks from contract summaries to custom agent development. The presentation, led by Google Cloud Legal Counsel Xavier Polidoro, gave law firms a practical look at tools that could keep more of their workflow inside Google's ecosystem.
What was on show
The demonstration covered three main areas. First, general prompting within Gemini Enterprise for accelerating contract summaries, flagging risks, and drafting preliminary client communications. Second, NotebookLM for grounded research - searching, cross-referencing, and summarizing dense legal repositories and multi-amendment agreements with full source grounding. Third, what Polidoro called Cloud Legal AI Helpers, custom chatbots that legal teams can build for deal management.
Polidoro described these as the "advanced" uses. The Helpers appear to connect to the recently announced Gemini Enterprise for Legal, though Google used different terminology during the demo. The company also mentioned that Harvey, Freshfields, and Inspira are using Gemini.
NotebookLM connects to massive document sets
NotebookLM stood out for its capacity. Users can add up to 300 sources, each with a 500,000-word limit. A law firm could connect the tool to a substantial collection of documents and data, then chat with those materials to extract insights. The product can also turn work product into an audio podcast - an unusual feature, though the core benefit is the ability to query across many sources at once.
Polidoro focused on how lawyers can "chat" with their documents and gain deep insights quickly. The source grounding means responses tie back to the original materials, which matters for legal work where provenance is everything.
Custom Helpers function like agents
Google offers two versions of the Cloud Legal AI Helpers: Pro and Flash. These connect to sources of the user's choice and can be customized for specific tasks. Polidoro showed an example where a Helper surfaced improved language from playbooks and trusted sources, placing it in a separate box on the page for the lawyer to cut and paste into a new document.
The terminology is still settling. Google called these "custom chatbots," but their ability to connect to chosen sources and perform directed tasks - redlining a document or helping write a brief - sounds closer to agents. The branding between Cloud Legal AI Helpers and Gemini Enterprise for Legal remains somewhat unclear, suggesting the product line is still being unified.
Google stressed that it does not train on any customer legal data.
Why this matters for legal professionals
Google's advantage here is centrality. Law firms already working within Google's platform can add legal-specific AI tools without leaving that environment. NotebookLM's 300-source capacity gives firms a way to query their own document repositories directly, rather than relying on a third-party legal database. The Helpers, while still early, let teams build task-specific tools without external vendors. The trade-off is that Google lacks its own legal data and polished point solutions. For now, firms must do the work of connecting sources and organizing agents themselves. The tools are useful, but not yet a fully joined-up legal platform.
Your membership also unlocks:
