Deployment guide

Self-host ToolJet on your infrastructure

Run the app builder, your apps and data connections in your own cloud account or on-premises. Operate the deployment yourself, or have ToolJet deploy and manage it in your AWS, Azure or GCP account.

Read deployment docs Discuss your deployment

Your cloud account
Self-managed or managed
On-premises and air-gapped

Overview

What self-hosting ToolJet means

A self-hosted low-code platform runs on infrastructure your organization controls. ToolJet's editor, application runtime and platform storage are deployed in your environment. Data connections reach the databases and APIs accessible from that environment.

Hosting location and operating responsibility are separate decisions. Your team can manage ToolJet, or the ToolJet team can manage a deployment in your own cloud account. AI gateway and model hosting are configured separately, as described below.

Operating model

Choose who manages your deployment

Both options keep the ToolJet platform in your infrastructure. Choose the operating model that fits your team's capacity and access policies.

Self-managed

Your team operates ToolJet

Install ToolJet on your own VMs or container platform. Your team configures networking, identity and data access, and manages monitoring, upgrades, backups and recovery.

Explore setup guides

Managed hosting

ToolJet operates it in your cloud account

ToolJet deploys and manages the platform in your organization's AWS, Azure or GCP account. The deployment stays in your account, with the access and operating arrangements agreed with your team.

Discuss managed hosting

For managed hosting, agree the deployment architecture, administrative access, upgrade process, backup and recovery responsibilities, and support scope before rollout.

Architecture

What runs in your environment

These are the main parts of a self-hosted deployment. The exact topology depends on your chosen setup guide and enabled features.

ToolJet application

Containerized services provide the visual editor, application runtime and server-side queries. Users access the deployment through your configured URL.

PostgreSQL

Uses two distinct databases: one for platform configuration and app definitions, and one for ToolJet Database tables. They can share a PostgreSQL server; back up both.

Redis

Provides caching and session management. Production sizing should account for your expected traffic and workload.

Database and API connections

Server-side queries use the credentials and network access configured in your deployment. Give each connection only the permissions its apps need.

Identity and application access

Configure user access, roles and authentication for your organization. SSO and other governance capabilities depend on your plan.

Optional AI services

The AI gateway and model endpoint can be external or customer-hosted. Self-hosting the app platform alone does not make AI inference local.

Connect systems reachable from your deployment

PostgreSQL
MySQL
Oracle
SAP HANA
MongoDB
Snowflake
BigQuery
REST API

Integration guides

Installation

Evaluate locally, then plan production

The Docker evaluation image includes PostgreSQL. For production, choose a supported deployment guide, configure persistent storage and secrets, and size the application, database and cache separately.

Docker evaluation command

docker run \
  --name tooljet \
  --restart unless-stopped \
  -p 80:80 \
  --platform linux/amd64 \
  -v tooljet_data:/var/lib/postgresql/16/main \
  tooljet/try:ee-lts-latest

This command serves ToolJet on port 80 and stores its bundled PostgreSQL data in the named Docker volume. Use test data for evaluation. Follow the production setup and upgrade guides before exposing a deployment to users.

Evaluation instructions

Infrastructure requirements

The documented starting sizes below apply to separate components, not the total deployment. ToolJet requires x86-64 Linux; arm64 is not supported. PostgreSQL 16.x is recommended and Redis 7.x is required.

Component	CPU	Memory	Storage
ToolJet application host	1 vCPU	4 GB	8 GiB
PostgreSQL	2 vCPUs	8 GB	20 GB
Redis	1 vCPU	512 MB	1 GB

Redis recommends 1 GB RAM for production. Size storage, replicas and capacity for your workload, enabled features and recovery requirements.

Full system requirements

AI hosting

Choose managed AI, BYOK or private inference

Self-hosting controls where ToolJet runs. Gateway and model hosting determine where AI requests are processed. Your deployment initiates key-authenticated outbound connections to ToolJet's hosted gateway; the gateway cannot initiate inbound access. Requests may include prompts, app context, schemas and data read through connected credentials.

Managed AI gateway

Use ToolJet-hosted AI

The default AI setup sends requests from your deployment to ToolJet's managed gateway and supported model providers. Allow outbound HTTPS to api-gateway.tooljet.ai and python-server.tooljet.ai, and limit the data available to AI.

Enterprise (Cloud or Self-hosted)

Bring your own model keys

BYOK lets you use supported model-provider keys through ToolJet's managed AI server without consuming ToolJet AI credits. It changes the credentials used for inference; by itself it does not move the gateway or model into your network.

Self-hosted Enterprise add-on

Host the AI gateway yourself

Deploy the gateway in your infrastructure and connect a supported model endpoint. To keep inference inside your network, the model endpoint must also be hosted there. An external model provider still receives requests.

For fully internal AI, host both the gateway and a supported model endpoint inside your network. BYOK with an external provider still sends requests to that provider. You can enable AI only in development and use test or sanitized data. Managed hosting and AI hosting are separate choices.

AI deployment options AI data usage policy

Security

Configure access and isolation for your deployment

Infrastructure ownership gives you control over network boundaries. Configure the platform's authentication, permissions and audit capabilities for your workload and licensed plan.

Identity

Connect SAML, OpenID Connect or LDAP on Team and above. Enterprise adds SCIM provisioning. Keep administrative access scoped to the people operating the platform.

Permissions

Set roles and access to apps, pages, components and data sources. Review both application permissions and the underlying database or API permissions.

Audit

Use audit logs to review platform activity. Confirm the retention, export and monitoring requirements for your plan and deployment.

Credentials

Data-source credentials use AES-256-GCM encryption. Protect encryption keys and environment secrets as part of your deployment and recovery procedures.

Security and AI controls Trust Center

Release planning

Plan the rollout before production

01
Choose infrastructure and an operating model
Decide where ToolJet will run, who will operate it, and how users and private data sources will reach it. For managed hosting, agree cloud-account access and responsibilities with ToolJet.
02
Configure identity and connections
Set up HTTPS, authentication and roles. Use separate credentials for each environment and least-privilege access to databases and APIs.
03
Test releases in an isolated environment
Use development and staging deployments to review apps, permissions and upgrades. Enterprise includes multiple deployments; GitSync promotes app definitions. Configure workspace constants and environment credentials separately.
04
Verify recovery and production access
Test database and persistent-storage restores, monitoring and incident ownership. Keep production credentials separate, and choose whether AI is enabled for each deployment.

GitSync Multi-instance configuration

Day-to-day operations

Keep the deployment maintainable

For self-managed hosting, your team owns these tasks. For managed hosting, agree how they are handled with ToolJet before launch.

Upgrades
Choose a supported release, read the migration guide and test upgrades in staging. Pin the version used in production rather than treating a moving image tag as a release policy.
Backups and recovery
Back up both PostgreSQL databases and required persistent storage. Preserve encryption keys and deployment secrets securely so restores can recover credentials. GitSync stores app definitions; it does not replace instance or data backups.
Monitoring
Monitor the application, database and Redis, and review logs and capacity. Establish who receives alerts and handles incidents.
Network access
Review outbound access for your chosen AI, identity, Git, email and data services. Environment variables can disable optional telemetry and update checks.

Upgrade guide Configuration reference Observability

Editions and licensing

Check the capabilities your deployment needs

Deployment requirements and licensed features are separate. Confirm your plan and any add-ons before designing the rollout.

Community and Free

The Community Edition is open source under AGPL-3.0. You can start with a free self-hosted deployment. Review the current Free-plan allowances and the edition that matches your requirements.

Team and Enterprise

Advanced identity, audit and release-management capabilities are licensed features. Enterprise includes multiple deployments; check the pricing page for the current feature breakdown.

Air-gapped Enterprise add-on

The air-gapped add-on supports operation without internet access, including offline licensing. Required data sources and supporting services must be accessible inside the isolated network.

Self-hosted pricing License configuration

Frequently asked questions

When should an enterprise self-host its internal-app platform?

Self-host ToolJet when your internal apps need access to private databases and APIs, your organization requires control over platform infrastructure, or your team needs deployment isolation. Your team can operate ToolJet, or agree managed hosting in your AWS, Azure or GCP account. Choose plan-specific identity and audit controls, then configure networking, backups and release processes. Hosting the platform yourself does not automatically make AI inference local.

What is self-hosted ToolJet?

Self-hosted ToolJet runs the editor, application runtime and platform storage in infrastructure your organization controls. Your deployment connects to databases and APIs reachable from that network. AI gateway and model hosting are configured separately.

Can ToolJet manage hosting in our own cloud account?

Yes. ToolJet can deploy and manage the platform in your organization's AWS, Azure or GCP account. Agree the architecture, administrative access, upgrade process, backup and recovery responsibilities, and support scope with the ToolJet team before rollout.

Can I self-host ToolJet for free?

Yes. A free self-hosted option is available, and the Community Edition is open source under AGPL-3.0. Advanced governance features and Enterprise add-ons are licensed separately. Refer to pricing for current plan allowances; managed hosting arrangements are discussed with the ToolJet team.

What infrastructure does ToolJet require?

ToolJet's documented requirements use x86-64 Linux, PostgreSQL 16.x recommended and Redis 7.x required. The application host starts at 1 vCPU, 4 GB RAM and 8 GiB storage. PostgreSQL and Redis have separate requirements, and production sizing depends on workload. Arm64 is not supported.

Can ToolJet connect to databases in a private network?

Yes, when your deployment has the required network access and credentials. Server-side queries can reach private databases and APIs without making those systems public. Configure least-privilege credentials and application permissions.

Does self-hosting keep all AI processing inside our network?

Not automatically. The default setup uses ToolJet's managed AI gateway. BYOK is available on Enterprise plans for both Cloud and Self-hosted. A customer-hosted AI gateway is a separate Self-hosted Enterprise add-on. Fully internal inference also requires a supported model endpoint inside your network; an external model provider still receives requests.

Can we run ToolJet without internet access?

Yes, with the Enterprise air-gapped add-on, including offline licensing. Keep required data sources, identity services and other dependencies inside the isolated network. If AI is enabled, internal inference requires a customer-hosted gateway and a supported local model endpoint.

Who handles upgrades and backups?

Your team handles them on a self-managed deployment. With managed hosting, agree the operating responsibilities with ToolJet. Test upgrades and restores before production changes. GitSync promotes app definitions and does not replace database backups.

Can we isolate development, staging and production?

Yes. Enterprise plans include multiple deployments. Use separate instances and credentials for development, staging and production, and promote app definitions with GitSync. AI can be enabled for development and disabled for the other deployments.

Plan your ToolJet deployment

Use the setup guides for a self-managed installation, or discuss managed hosting in your cloud account with the ToolJet team.

Deployment docs Discuss deployment

Page updated 16 Oct 2026