Enterprise Deployment

​

Overview

Devin has a straightforward setup. He can be awoken from the website, Slack, or the API.

When he is awoken, Devin is now ready to help solve your problems. No matter how simple or difficult his task is, Devin does not change.

Devin has a workspace, which includes a shell, a browser, and code-editor. Devin navigates his workspace under direction from his brain.

Devin’s workspace can be deployed in two ways.

• SaaS (Multi-Tenant)
• Virtual Private Cloud (VPC) (Single-Tenant)

Devin’s brain always resides in Cognition’s Cloud.

Deployment	Primary Advantage	Disadvantage
SaaS	Fast Setup	Multi-Tenant
VPC	Isolated	Slower Deployment

Deploying Devin via SaaS allows you get up and running in hours.

However, many customers opt to deploy Devin in their virtual private cloud (VPC) due to IT, Security, and compliance requirements.

After deciding on a deployment model, the next most important step is connecting your source code.

We support several code repository integrations

• Github
• Gitlab
• Bitbucket
• Github Enterprise

​

Devin’s Architecture

​

VPC Deployment Specifications

Devin’s workspace has a few requirements for the server it will utilize.

In order to set up Devin’s workspace, we will install a hypervisor on the provisioned server. This means that the server must support virtualization — Devin does not run in containers.

​

Cross Tenant Communication

Feature	Requirement
Networking	Egress
Ports	HTTPS/443
Connection	On boot, a websocket opens and connects to an isolated container in the Devin tenant
Communication	All subsequent exchanges happen over this connection
Isolation	Backend Session Isolation