Encore vs Void: Where Should Your AI-Built Internal Apps Run?

Cloudflare open-sourced Void in October 2026 with a pitch aimed at the same problem many teams have with internal tools: employees build apps with coding agents, and IT does not want to hand cloud access to every one of them.

The New Stack's coverage describes the three parts as an SDK, a CLI that provisions Cloudflare resources on deploy, and a platform companies host to manage those apps. Encore solves an overlapping problem with a different runtime and a different cloud.

Encore vs Void at a glance

Void: Vite apps on a self-hosted Cloudflare platform

Void comes from VoidZero, the team behind Vite, which Cloudflare acquired. You add it as a Vite plugin and write server routes and pages in the same project. Importing void/db gives the app a local database in development, and void deploy provisions D1 in production and applies migrations. Projects can instead use PostgreSQL or MySQL that you host, connected through Hyperdrive. Imports for AI, queues and other resources work the same way, and the CLI records resource IDs in a lock file you commit.

The platform is the part aimed at internal tools. A company installs it in its own Cloudflare account on Workers for Platforms with R2, a PostgreSQL database for platform state and a login provider. Employees run void connect against it and deploy with the same command, while admins manage users, project access, resource limits, logs and rollbacks. Employees do not need their own Cloudflare credentials to ship.

Good to know

Apps run on the Workers runtime, so Node APIs and long-running processes follow Workers' rules. The default D1 is built on SQLite, which the SQLite project describes as a strong fit for many workloads with fewer concurrent writers than a client-server database. If your internal data is in Postgres on AWS, the app reaches it through Hyperdrive, either over the public internet or through a tunnel you set up.

Encore: backends in your own AWS account

Encore combines an open source SDK for TypeScript and Go with a platform that provisions and deploys each service's infrastructure into your own cloud account. A service declares its database, Pub/Sub topics, cron jobs and secrets in code, and the platform creates them on AWS, for example RDS for databases and SNS with SQS for topics, inside a VPC it sets up per environment.

// refunds/refunds.ts
import { api } from "encore.dev/api";
import { SQLDatabase } from "encore.dev/storage/sqldb";
import { Topic } from "encore.dev/pubsub";

const db = new SQLDatabase("refunds", { migrations: "./migrations" });

export const refundRequested = new Topic<{ refundID: number }>("refund-requested", {
  deliveryGuarantee: "at-least-once",
});

export const request = api(
  { expose: true, auth: true, method: "POST", path: "/refunds" },
  async (p: { orderID: number; reason: string }): Promise<{ id: number }> => {
    const row = await db.queryRow<{ id: number }>`
      INSERT INTO refunds (order_id, reason) VALUES (${p.orderID}, ${p.reason}) RETURNING id`;
    await refundRequested.publish({ refundID: row!.id });
    return { id: row!.id };
  },
);

encore run starts Postgres locally and runs the topic and its subscribers on your machine, and the agent can inspect traces and schemas through the local MCP server. An agent that adds the topic above has also declared the queue behind it and can run the change locally before review.

On the governance side, an Admin can require manual infrastructure approval per environment, and only Admins can approve it.

Roles decide who can deploy and who can change cloud connections. On the Encore Pro plan, pull requests get preview environments with their own databases. The agent pushes code, and cloud access sits with Encore's IAM role, not the agent.

Good to know

Encore covers the backend, and the frontend comes from you, either served with api.static or deployed to a frontend host. Custom roles beyond Admin, Member and Viewer are an add-on to the Encore Pro plan.

How to choose

• Encore when internal tools need to sit next to your data in AWS or GCP and go through reviewed infrastructure changes.
• Void when your company runs on Cloudflare and wants a self-hosted portal for employee-built Vite apps.

Related guides

• Retool alternatives for teams that build internal tools with AI agents
• Building internal tools with AI agents without giving them your AWS keys
• How AI agents deploy with Encore on AWS
• Best infrastructure for AI coding agents
• How to give AI coding agents access to cloud infrastructure safely
• How to test AI agent pull requests in a preview environment