Enterprises can now manage autonomous software sprawl and rising operating costs through a single governance platform. Thoughtworks has launched Agent/works, giving organisations one control plane and a governed runtime deployable across any cloud.
The platform targets a sharp shift from early AI experimentation to high-stakes operational deployment. Thoughtworks unveiled it at the Databricks Data + AI Summit in San Francisco, partnering with Databricks on enterprise AI governance.
According to Sonar's 2026 State of Code Developer Survey, 42 per cent of committed code is now AI-generated or AI-assisted. Separate research from AppSec Santa found 25 per cent of AI-generated code samples contained critical security vulnerabilities.
Unmonitored autonomous systems risk exposing sensitive data, breaching compliance rules and obscuring fast-growing AI operating costs.
Agent/works analyses every path through a workflow and confirms a fully compliant route exists before an agent runs. It grants capability-based, scope-bound and time-limited permissions that narrow automatically as agents touch sensitive systems.
The governed runtime applies adaptive policies during execution, covering both end-to-end workflow agents and interactive coding tools.
A centralised registry provides visibility, evaluations, usage analytics and cost controls across every agent, model, tool and policy.
The platform runs on a multi-model backend, registering any model through a standard API and connecting third-party tools and services. This architecture lets enterprises scale automation safely without governance becoming a bottleneck.