Encore is an infrastructure automation platform for developers and AI agents, covering the entire process from development to production, designed to eliminate the bottleneck of waiting for Terraform PRs or platform tickets, enabling developers and agents to quickly build, validate, and ship applications.
Key capabilities:
- Declare infrastructure directly in application code: Developers can define the databases, queues, buckets, and other resources required by features directly in code. These resources run alongside the application in the local development environment, allowing complete features to be built, run, and validated without deploying to real cloud resources.
- Validate application and infrastructure changes in the local loop: Encore allows application and infrastructure changes to be validated in the local development loop without deploying to real cloud resources. When ready to release, Encore automatically provisions the required infrastructure in the user's own cloud account.
- Preview environments: After pushing a branch, Encore creates a dedicated temporary environment that can integrate with external systems and validate complete workflows, avoiding shared staging environments where changes by others break testing.
- Provision infrastructure in the user's cloud at deployment: At deployment, Encore automatically provisions resources such as databases, queues, and buckets in the user's AWS or GCP account. Users retain their own cloud account, VPC, and compliance boundaries.
- Centralized configuration and self-service: Environment-specific configuration is managed centrally and separated from the development loop. Developers and agents can operate independently within approved policies, without every change going through platform review.
- Progressive adoption: You can start with a single service, team, or new project. Encore coexists with your existing tech stack, deploys into your VPC, and scales gradually.
- Cloud sandboxes (coming soon): Each agent gets an independent environment containing the application and infrastructure, running independently of the local machine, and can start multiple parallel workstreams on demand.
How to use:
- Run the installation command in the terminal to get started locally. After installing Encore, you can use an AI agent to build, run, and deploy an application with real infrastructure in minutes.
- Developers write application code, and Encore automatically sets up the required cloud resources, including networking, IAM, and secrets, all created in the user's own account and inspectable at any time in the cloud console.
- There is no need to write separate configuration files or maintain Terraform configurations or Pulumi stacks. Infrastructure is declared directly in application code, and Encore automatically provisions it in each environment.
- Agents define infrastructure in the code they write and run and validate the complete application locally or in preview environments; Encore's MCP server provides agents withProvides context about services, APIs, and tracing.
Applicable scenarios:
- Teams that want to adopt infrastructure automation without rewriting their platform can start with a new service or new project, coexisting with existing infrastructure and the Terraform that manages it.
- Teams that need end-to-end test environments for each feature can use dedicated preview environments to validate complete workflows.
- Teams developing with AI agents can have agents work in parallel in isolated environments without touching cloud credentials.
- Teams concerned about vendor lock-in, because deployments go to their own cloud accounts, production traffic does not pass through the Encore platform, there are no runtime dependencies, they can deactivate the Encore platform at any time while retaining existing infrastructure, and use Encore open-source tools to build their own CI/CD.
Limitations and support:
- Encore supports TypeScript and Go, both with the full feature set; Python and Rust support is coming soon.
- Deployment targets are the user's own accounts on AWS or GCP; the frontend can use any framework, and a typed API client is generated for the backend.
- Agents do not obtain cloud credentials, and infrastructure declarations are provisioned deterministically by Encore, following security and reliability best practices; environment-specific configuration such as instance sizes and secrets is applied in the Encore control plane, separate from the code written by agents. Each pull request gets an isolated preview environment, and changes have an audit trail.
- Permissions are implemented through IAM roles (AWS) or service accounts (GCP) granted by the user in their own account and can be revoked at any time; Encore uses them to create and update resources and will not delete infrastructure without approval. The enterprise edition supports just-in-time permissions granted only when specific changes are approved.
- Billing: The Starter plan is free for local development and small projects; the Pro plan is $49 per team member per month, plus per-environment and per-resource fees. Cloud resources are billed directly by AWS or GCP in the user's account. Cloud bills and credits are used as usual, and credits cover fees paid to AWS or GCP, not Encore plan fees.
- No vendor lock-in: deploy to your own cloud account and own and control the infrastructure. If you stop using the Encore platform, you can retain existing infrastructure and use Encore open-source tools to build your own CI/CD pipeline.
- Production environments: Encore provisions standard managed services in the user's own cloud account, such as RDS, Cloud SQL, ECS on Fargate, Cloud Run, and GKE; the application runs in them, with no runtime dependency on the Encore platform, production traffic does not pass through Encore, and there is no need to migrate as it grows. Teams such as Groupon and Coinbase run Encore workloads in production.
- Adoption approach: Services can be converted to the Encore SDK one by one, or existing databases, buckets, and topics can be connected rather than recreated; any AWS or GCP service can be called directly in code and coexist with Encore-managed resources.
Other information: Official channels are provided for community discussion, project contributions, following on X, and tutorial videos.
