ToolJet review: the open-source Retool alternative for Next.js internal tools (2026)

ToolJet keeps showing up in every "open-source Retool alternative" thread, and in 2026 it wears a new label: ToolJet AI. We run our own stack on Next.js and self-host what we can, so the question we actually care about is narrower than "is ToolJet any good." It is this: does ToolJet earn a place next to a Next.js SaaS, and what do you really get for free? We spun it up, read the license, and priced out the AI. Here is the honest read.

Quick answer (August 2026)

ToolJet is a self-hosted, open-source (AGPL-3.0) low-code builder for internal tools such as admin panels, back-office dashboards, and ops consoles. You run it as its own service and point it at your existing database. It is one of the strongest Retool alternatives when your requirement is "self-host it, keep the data in-house, build faster than a hand-rolled React admin." Two caveats decide the call for a Next.js team. First, ToolJet does not compile into your Next.js app or run on Vercel serverless; it is a separate Node and React server you host beside your app. Second, the genuinely free path is the self-hosted Community Edition (a limited feature set), while the headline AI generation runs on metered credits that are thin on the free cloud tier. If you want an internal tool that lives inside your product's codebase, a code-first library fits better. If you want a standalone ops UI over your Postgres, ToolJet is a real answer.

The 30-second version

What ToolJet actually is

ToolJet is an open-source low-code platform for building internal tools: a visual drag-and-drop app builder, a broad set of data connectors (databases, REST and GraphQL APIs, cloud services), a built-in datastore called ToolJet Database, and multi-step queries and workflows. The repo (ToolJet/ToolJet, roughly 38.3k GitHub stars as of August 2026, AGPL-3.0) is a JavaScript and TypeScript project on Node.js and React. In 2026 the project brands itself as "the open-source foundation of ToolJet AI," an AI-native platform for building internal tools, workflows, and agents.

The mental model that matters: ToolJet is a server you host, not a component you import. You deploy it (Docker, Kubernetes, Helm, or a managed cloud), give it access to your data sources, and your team builds apps inside its editor. Those apps are served by ToolJet, behind ToolJet's auth, at ToolJet's URL. They do not live inside your Next.js routes.

The Next.js and Vercel fit lens

This is the part most round-ups skip, and it is the one that decides the tool for our readers. ToolJet is a persistent Node and React application with its own PostgreSQL database (and a Redis cache). That has two consequences.

• It does not deploy to Vercel serverless, and it does not become part of your Next.js build. You run it as a sibling service, in its own container or VM, next to the app.
• You integrate at the data and auth layer, not the code layer. ToolJet talks to the same database and APIs your Next.js app uses, and you gate it behind SSO. If you wanted the tool to live inside your product (same repo, same deploy, same design system), a code-first admin library like Refine is the closer fit, because it compiles into the Next.js app. ToolJet is the better answer when the internal tool is a separate back-office surface your ops team uses, not a customer-facing screen.

What you actually get for free (the CE vs AI reality)

"Open source" and "free" are not the same line here, and the 2026 AI pivot muddies it, so let us be precise.

• Self-hosted Community Edition: free, open-source, AGPL-3.0. You host it and you get the visual builder, the connectors, and the datastore. ToolJet describes the CE as a limited feature set relative to the paid tiers; the richest AI and enterprise-governance features sit upstream.
• ToolJet Cloud (hosted): AI generation (App Generation, Query Generation, Auto-fix and Debugging) is offered across the tiers, but it is metered by AI credits, per ToolJet's pricing (2026):
  • Free, $0/mo: 2 builders, 50 end users, 2 apps, and 100 AI credits per month with no add-on credits.
  • Pro, $79/mo: unlimited builders, 50 end users, 5 apps, 2,000 AI credits per builder per month.
  • Team, $199/mo: unlimited builders, 100 end users, unlimited apps, 3,000 AI credits per builder per month.
  • Enterprise, from $3,000/mo: unlimited users and apps, custom credit allocation.

The honest read: the free cloud tier's AI is a demo, not a daily driver. One hundred credits with no way to top up means you will try the AI builder, like it, and hit the wall inside a real project. If AI generation is the reason you are here, you are pricing the Pro or Team tier, or you are bringing your own model into the self-hosted CE. If you just want a hand-built admin panel without the AI, the self-hosted CE is genuinely free and does the job.

The license question nobody answers

ToolJet is AGPL-3.0, and that matters more than the usual "it's open source, ship it" hand-wave.

• An internal admin panel behind your firewall: AGPL is a non-issue. You use it, you do not distribute a modified version, and you are done.
• You fork ToolJet, modify it, and expose the modified version to users over a network: AGPL's network-copyleft clause obligates you to offer those users your modified source. For most teams building internal tools this never triggers. But if your plan was to fork ToolJet into a customer-facing product, read the license first, because that is the exact scenario AGPL is written for.

Compare that to the code-first tier: Refine is MIT and Appsmith is Apache-2.0, both more permissive if embedding-in-a-product is on your roadmap. We put ToolJet head-to-head with the other GUI builders in our Appsmith vs ToolJet vs Budibase breakdown.

Lock-in: JSON apps, not portable code

ToolJet apps are stored as JSON app definitions that run inside the ToolJet runtime. You are not generating a React codebase you can lift out and own; you are committing to ToolJet as the host. That is the standard trade for every GUI internal-tool builder (Retool, Appsmith, and Budibase all do a version of this), and it is fine as long as you know it going in. It is also the cleanest reason a code-first library wins when "we own every line" is a hard requirement.

Where ToolJet wins, and where it does not

Choose ToolJet when:

• You want a self-hosted internal-tools platform and data residency matters.
• Your builders are technical enough for a low-code editor, but you do not want them writing a React admin from scratch.
• You need to wire many data sources (Postgres, REST, GraphQL, cloud services) into one back-office UI fast.

Look elsewhere when:

• You want the tool inside your Next.js app and repo, which points to a code-first stack (Refine or React-Admin).
• You want the biggest widget library and the most advanced GUI capabilities; Appsmith is the heavier GUI builder, and ToolJet trades some breadth for a cleaner editor.
• You want no-code aimed at business users rather than developers; Budibase leans more that way.
• You are migrating off Retool specifically, in which case weigh the whole field in our open-source Retool alternatives guide.

Even ToolJet's competitors concede the shape of it: Refine's own comparison page frames ToolJet as working "well for basic internal tools with self-hosting requirements," which, stripped of the sales angle, is a fair description of its sweet spot.

The bottom line

ToolJet in 2026 is a solid, genuinely open-source, self-hostable internal-tool builder that earns its Retool-alternative reputation for back-office apps over your own data. For a Next.js SaaS team, hold two facts in your head before you commit. It runs as a separate service, not inside your app, and the free tier's AI is trial-sized, while the free value is the self-hosted Community Edition. Match those to your actual constraint (separate ops UI vs in-app tool, AI-first vs hand-built, self-host vs managed) and the pick makes itself.