The AI DevOps agent that deploys your repositories, with your approval
Give it your GitHub or GitLab repositories, or a .zip of the project. It reads the code, draws the architecture, proposes a priced plan, then deploys when you approve. And if the application does not exist yet, the built-in coding agent writes it with you.
From repository to production in four steps
You stay in control at every step: the agent proposes, you decide.
Give it the code
Pick the repositories that make up the system, even if it is split across several. No repository? Upload a .zip of the folder.
It draws the architecture
It finds the services and how they talk to each other, then draws the diagram in front of you. What is missing is flagged, not invented.
It proposes a priced plan
Each component, its size and its price. Nothing is created until you approve.
It deploys, within the plan
Databases, services, private network, HTTPS addresses. The platform refuses any action outside the approved plan.
An agent you can trust with production
Nothing without your approval
The plan is recorded, you approve it, and that control does not rely on the model behaving: the platform itself holds to it.
Your passwords stay out of its sight
It deploys with your passwords and keys without reading them. Secrets only you hold go straight to the vault, without passing through the conversation.
A whole system
Frontend, API, workers, databases and message queues are deployed as one, connected by a private network.
The diagram comes with it
You leave with a drawing of your architecture, taken from the code and editable.
It decides instead of interviewing you
Obvious choices are stated, not asked. When the code does not settle a question, it asks and recommends an option.
Everything traced, nothing kept
Every command it ran stays available to read. The code cloned for the deployment is deleted once the work is over.
The protections, and what stays in your hands, are described in the DevOps Agent documentation.
The built-in coding agent: from an idea to a live application
In AI Studio, describe what you want to build. The agent writes the code in an isolated environment, runs the application, reads the errors and fixes them, while you follow the live preview.
The two agents complement each other: one writes the application, the other puts an existing system into production.
What people ask about the agents
What is an AI DevOps agent?
It is an assistant that does the work of a deployment engineer: it reads a project’s code, works out which services it is made of, prepares the infrastructure it needs and puts the application online. ISOGrid’s does this from your GitHub or GitLab repositories, or from a .zip, and deploys nothing without your approval.
Can the agent deploy without my approval?
No. It proposes a plan listing each component, its size and its price. Nothing is created before you approve, and the platform then refuses any action outside the approved plan.
Does the agent see my passwords?
It deploys with your passwords and keys without reading them, and the secrets you give it on its card go straight to the vault. Do keep secrets out of your repository, though: the agent reads your files to understand what it deploys. The documentation details what is protected and what stays in your hands.
What happens to my code after the deployment?
The code cloned for the deployment is deleted once the work is over. The applications, databases and services that were deployed stay in place, as do the diagram, the plan and the conversation.
What is the difference between the DevOps Agent and the coding agent?
The coding agent writes an application with you in AI Studio. The DevOps Agent puts an existing system into production without changing its code. You can use either one, or both.
What does it cost to try?
Registration is free and every new account receives free tokens to try the agents. Before any deployment, the plan shows the price of each component.
Read next
Let's talk about your project
Write to us or call us. We answer in French, Arabic and English.
