How a legal delivery firm turned 1,100 staff into AI builders

ABC Legal, a U.S. legal document delivery company, has quietly become one of Anthropic's more revealing enterprise case studies. By July 2026, roughly 310 employees across every department, not just engineers, were using Claude for daily work, and together they had built more than 50 agents now running in production.

That scale followed a company-wide rollout of Claude Enterprise to about 1,100 employees in early 2026. Rather than centralizing automation inside an IT or data team, ABC Legal opened agent-building to the whole workforce, using Claude Managed Agents, Anthropic's deployment layer for shared workspaces, unified audit and billing, and always-on cloud-hosted agents.

The company's CTO, Brandon Fuller, oversaw an architecture that treats agents less like scripts and more like software. Agent configurations, including prompts, tool lists, schedules, credentials, and memory, live in git repositories alongside the rest of ABC Legal's codebase. That gives non-technical staff a version-controlled, auditable way to modify or extend agents without rebuilding them from scratch each time.

Two kinds of agents, one shared workspace

The 50-plus agents fall into two categories. Event-driven agents react to triggers such as a new job entering the system or a document being returned. Scheduled agents run on fixed timers, hourly, daily, or weekly, handling recurring administrative work without a human initiating each run.

That structure matters for a company whose core business, serving legal documents, involves high volumes of repetitive, time-sensitive paperwork spread across many teams. Giving frontline employees rather than a central engineering group the tools to automate their own workflows let the agent count grow quickly once the infrastructure was in place.

Measurable cost impact

The clearest evidence of impact is financial. Before heavy optimization, some agents cut the cost of the human tasks they covered by up to roughly 50%.

Anthropic has increasingly used case studies like this one to argue that enterprise AI adoption works best when deployed broadly rather than confined to specialist teams. ABC Legal's numbers, an 82% adoption rate of Claude Enterprise translating into over 50 self-built production agents, offer one of the more concrete data points supporting that pitch, though it remains a single vendor-published account rather than independently verified financial disclosure.