Gemini Cloud Assist is an AI assistant from Google built into the Google Cloud console, opened from a “Cloud Assist Chat” control at the top right of the console. The source available here is a Google Codelabs workshop, “AI Agent End to End”, which uses it at the two ends of the software development lifecycle: first to turn an idea for an application into a cloud architecture, and later as an operational assistant for the running application.
Capabilities
In the workshop’s design step, a user describes the application they want in a prompt — in the example, a Python backend and React frontend hosted separately on Cloud Run, communicating over WebSocket, with generated images stored in a Cloud Storage bucket and a way to store an API key — and Cloud Assist generates an architecture diagram. From an “Edit app design” view the user can then download the corresponding Terraform code, which the workshop describes as the machine-readable version of the same plan. The workshop treats that code as a reference blueprint rather than something the participant runs.
In the operations step, after the application is deployed, the workshop uses Cloud Assist on the live Cloud Run service with prompts for three kinds of task: summarising recent errors in the service’s logs, investigating high startup latency, and analysing the costs of the service and its storage bucket for savings. Enabling it is shown as a “Get Gemini Assist” step with an option to enable Cloud Assist at no cost.
Adoption & Ecosystem
The workshop pairs Cloud Assist with Gemini CLI, which it uses for writing code, tests, deployment scripts and a CI/CD pipeline in between the design and operations steps, and with Agent Development Kit (ADK) for the agents the application contains. Its framing is that the same AI assistant that helped build an application is also a partner for monitoring, troubleshooting and optimising it in production. This is Google presenting its own tools in a training workshop, not an independent account of how they are used.