Internal Tools 2026 — Retool vs Budibase vs ToolJet

Every growing company eventually accumulates internal software nobody planned for: an admin panel to refund a customer, a dashboard to see which orders are stuck, a queue where support reviews flagged accounts. Engineers can build these by hand, but each one is a small web application with a login, permissions, a database connection, and a maintenance bill. Internal tool builders exist to compress that work. This guide explains what they are, compares Retool, Budibase, ToolJet, Appsmith, and Plasmic using public pricing and license information, and lays out how to choose.

What an internal tool builder actually is

An internal tool builder is software for assembling applications that only your own staff use. You drag components onto a canvas (tables, forms, buttons, charts), connect them to data sources (a Postgres database, a REST API, a spreadsheet), and write small queries and scripts that say what happens when someone clicks. The platform supplies the parts that are tedious to rebuild every time: authentication, role-based access, hosting, and a library of widgets.

A fair analogy is a prefab kitchen. A custom-built kitchen fits perfectly but takes months. A prefab one arrives in boxes, covers 90% of what you need, and goes up in a week, though you cannot move the plumbing. Builders work the same way: fast to assemble, constrained at the edges.

The category sits between neighbors:

• No-code app builders such as Bubble or Glide target customer-facing apps and non-developers.
• Business intelligence tools such as Looker or Metabase focus on reading and charting data. Our Tableau vs Power BI vs Looker comparison covers that side. Internal tools add the ability to write back: refund an order, change a plan, approve a request.
• Workflow automation such as Zapier or Make runs tasks between apps without a user interface. The Zapier vs Make vs n8n breakdown covers that layer, and our no-code automation guide explains the ideas from scratch.
• Data app frameworks such as Streamlit turn Python scripts into interfaces. See our Streamlit tutorial.

The contenders at a glance

Pricing and licensing below come from each vendor's public pricing page and GitHub repository, checked on October 22, 2026. Prices change, and some pages show different figures for monthly and annual billing.

Tool	Model	License	Free tier
Retool	Proprietary cloud; self-hosting on all plans	Proprietary	Up to 5 users, unlimited apps, 500 workflow runs per month
Budibase	Open source plus paid cloud	GPL v3 (client and components under MPL; paid features under Business Source License)	Open source self-host: 1 workspace, unlimited apps and users
ToolJet	Open source plus paid cloud and enterprise	AGPL-3.0	2 builders, up to 50 end users, up to 10 apps
Appsmith	Open source plus paid cloud	Apache-2.0	Up to 5 cloud users; free Community edition
Plasmic	Visual builder for React and web content	Proprietary	Free plan listed

Sources: Retool pricing, Budibase pricing and repository, ToolJet pricing and repository, Appsmith pricing and repository, and Plasmic pricing.

The pattern is the usual one. Managed products trade money for time; open-source products trade time for ownership. Neither is wrong, but the trade should be made deliberately.

Retool: the polished commercial option

Retool is the best-known commercial product in the category. It is a proprietary platform where builders compose apps from a large component library, write JavaScript in query and transformation fields, and connect to databases and APIs. It also has a workflows feature for background jobs, which overlaps with automation tools.

What it costs

According to the pricing page, Retool charges by role. A builder is a user who created or edited an app or workflow during the billing cycle; an internal user is an enabled user who did not. On annual billing, the Team plan is $10 per builder and $5 per internal user per month ($12 and $7 on monthly billing). The Business plan is $50 per builder and $15 per internal user per month ($65 and $18 monthly), and adds external-user pricing, with the first 50 external users free. Enterprise is custom. Annual billing is stated to save 20% against monthly.

The free plan covers up to five users with unlimited web and mobile apps, 500 workflow runs per month, 5 GB of database capacity, and 5 GB of file storage. Retool lists self-hosting on all plans, with a 5-user limit on Free and 100 users on Team and Business, though its page is internally inconsistent about which features carry over to self-hosted deployments, so confirm the specifics with the vendor before depending on them.

Strengths and limits

Strengths. A deep component library, JavaScript available throughout, and mature access controls. A team that wants a working internal app quickly and has budget will reach one soonest here.

Limits. The role-based pricing means costs rise when many people build or edit apps in the same month. Apps live inside Retool's runtime, so leaving later means rebuilding, which is the practical meaning of lock-in. And because the editor is flexible, complex apps can turn into a tangle of queries that only their author understands.

Budibase: open source with a built-in database

Budibase positions itself as an open-source low-code platform you can run on your own infrastructure. The repository describes the project as licensed under GPL v3, with the client and component libraries under MPL, which it says means apps you build can be licensed however you like, and paid features under the Business Source License.

What it costs

The open-source edition is free to use indefinitely, with one workspace, unlimited apps, automations, and users, and SSO included. The cloud plans, per the pricing page, are Pro at $19 per month, Premium at $49 per month, and Business at $299 per month, with usage measured in "automation actions" per month that function as hard limits. End-user and creator add-ons cost $5 and $50 per user per month respectively. Enterprise pricing is custom.

Strengths and limits

Strengths. A free self-hosted tier with no user cap is hard to match. It can use its own database or connect to external ones, and it ships with SSO on the open-source tier.

Limits. GPL v3 is a copyleft license. For a team that merely runs the software internally, that rarely causes problems, but a company that wants to modify and redistribute it should read the license terms with counsel. The component library is smaller than Retool's, and the cloud plans meter automation actions, so a busy workflow can hit a ceiling.

ToolJet: modern open source with an AGPL license

ToolJet is open source under the AGPL-3.0 license. The AGPL differs from the GPL in one important way: it extends copyleft to software accessed over a network. If you modify AGPL software and let users interact with it over a network, you must offer them the source of your modified version. Running ToolJet unmodified for your own staff is generally straightforward, but modifying it and exposing it to outside users is the situation to check with a lawyer first.

What it costs

The free plan allows two builders, up to 50 end users, up to 10 apps, 100 AI credits per month, 10 ToolJet database tables with 500 rows, and community support via Slack. Paid cloud plans on the pricing page are Basic at $23, Pro at $79, and Team at $199 per builder per month, with Enterprise at custom prices and no per-user fees. The page notes the Team price is the yearly rate, with $249 listed for monthly billing, and does not state the billing period for Basic and Pro.

Strengths and limits

Strengths. Many self-hosting options are documented, including Docker, Kubernetes, Helm, and major cloud providers. The interface is contemporary, and the free plan has a generous end-user allowance.

Limits. The free tier's limit of two builders and 500 rows in the built-in database is tight for anything beyond evaluation. Per-builder pricing climbs quickly as the number of app authors grows.

Appsmith: the Apache-licensed veteran

Appsmith is open source under the Apache-2.0 license, which is permissive. Unlike GPL or AGPL, Apache-2.0 does not require you to release your modifications, which makes legal review simpler for companies wary of copyleft.

What it costs

Per its pricing page, the Free plan covers up to five cloud users, five workspaces, Git-based version control limited to three repositories, Google SSO, and community support. Business is $15 per user per month. Enterprise is listed at $2,500 per month flat for 100 users. The Community edition is the free open-source platform for self-hosting, and the page lists airgapped and managed hosting as Enterprise add-ons. It does not state clearly which paid plans include self-hosting.

Strengths and limits

Strengths. A permissive license, Git integration even on the free plan, and a JavaScript-first scripting model.

Limits. Its pricing is per user, so a company with many occasional users pays for each one. Features such as audit logs and advanced access control sit in the paid tiers.

Plasmic: a different kind of tool

Plasmic is a visual builder aimed at designers and developers working with React and web content. It is not a direct competitor to Retool in the "wire a table to a database" sense. Its pricing page lists Free, Starter at $39, Pro at $103, and Scale at $399 per month billed yearly, plus Enterprise. A "code generation" row appears in the comparison table, but the page text does not show which tiers include it, so check the documentation before assuming a specific export workflow.

A reasonable use case is a team that wants internal pages to share a design system with its customer-facing app, and that has React developers to wire up data. For standard admin panels, the purpose-built tools above save more time.

Cost: a worked comparison for ten users

Comparing plans is hard because each vendor counts something different: builders, users, actions, or flat tiers. The only fair approach is to price a specific team. Take ten people, two of whom build and edit apps and eight of whom only use them, on annual billing. The figures use list prices from the pages above and ignore taxes, overages, and enterprise discounts.

Platform	Calculation	Approximate monthly cost
Retool Team	2 builders at $10 plus 8 internal users at $5	$60
Retool Business	2 builders at $50 plus 8 internal users at $15	$220
Appsmith Business	10 users at $15	$150
Budibase open source, self-hosted	Software free; you pay for hosting and maintenance	$0 plus infrastructure
ToolJet Free	2 builders, 8 end users within the 50-user cap	$0, within free-tier limits

Read the table as an illustration, not a quote. Retool's free plan covers five users, so a five-person team might pay nothing. ToolJet's free plan covers two builders and ten apps, which a growing team will exceed. Budibase's "free" has a real cost in the engineer-hours to run, patch, and back it up.

The honest accounting for self-hosting includes a server or container service, database backups, upgrades, monitoring, and someone on call when it breaks. A team with an existing platform group may absorb that cheaply. A team of three with no operations experience may find a $60 monthly bill the better deal.

Where internal tool builders earn their keep

Admin panels for products. Look up a user, change a plan, reset a flag, issue a refund. Almost every software company needs this, and it rarely deserves custom engineering.

Operations dashboards. Stuck orders, fulfillment queues, support backlogs. The value is combining several data sources on one screen with actions attached.

Lightweight CRMs and trackers. When a spreadsheet has outgrown itself but a full CRM is overkill. Our article on why spreadsheet automations break at scale describes the stage at which teams usually make this move.

Review and approval queues. Content moderation, expense approvals, and similar workflows where a person must look at an item and click approve or reject. Our piece on how approval workflows prevent automation mistakes explains why a human checkpoint is often the safest design.

Data labeling and annotation interfaces. Machine learning teams build small interfaces for reviewers to tag examples.

These tools are a poor fit for high-traffic public applications. They are optimized for staff-scale usage, and their pricing and architecture assume that.

Build, buy, or assemble: when a builder stops paying off

The decision is easier when you separate three costs: the cost to create the first version, the cost to change it, and the cost to leave. Builders are strongest on the first, reasonable on the second, and weakest on the third.

Creating the first version. A builder removes a large share of the setup work: login, hosting, permissions, a table component with sorting and filtering, and a form wired to an update query. A custom framework app needs each of those decided and built. For a screen that lists records and lets someone edit one, the difference can be days versus hours.

Changing it. Early changes are cheap in a builder because you rearrange components. Later changes get harder when logic accumulates in dozens of query fields and event handlers spread across pages. Code in a repository has tooling that builders often lack: code review, automated tests, search, and refactoring. A rough signal that you have hit this wall is when every edit makes someone nervous about what else will break.

Leaving. Apps built on a proprietary runtime generally cannot be exported as working code. Open-source platforms reduce the risk because you can keep running the software even if the vendor changes terms, though your apps still depend on that platform's format. Plasmic's code generation, where available, is aimed at the opposite situation, producing React code you own.

A practical rule is to start in a builder for anything a handful of staff will use, keep the business logic that truly matters (pricing rules, eligibility checks) in your own backend behind an API, and let the builder call it. That way the screens are disposable and the logic is not. If you later move platforms, you rebuild interfaces, not rules.

Teams also underestimate the value of a short written spec. Before opening any builder, list the people who will use the app, the three actions they must perform, the data each action reads and writes, and what should never be possible. That one page prevents most rework.

Security and governance: the part that gets skipped

An internal tool is a window into your production data, often with write access. That deserves more care than the speed of building suggests.

Credentials. Connect to databases with a dedicated account that has only the permissions the app needs. A reporting tool should not hold a superuser password. Store secrets in the platform's secret management, not in query text.

Network access. Databases should not be exposed to the open internet for the sake of a tool. Use a VPN, SSH tunnel, private networking, or an IP allowlist so only the platform can connect.

Roles. Decide who can view, edit, and administer each app. Test with a low-privilege account, not just your own admin login.

Audit trail. For anything that changes money or customer records, you want a log of who did what. Audit logging is often a paid-tier feature (Appsmith lists it under Business), so factor it into the comparison.

Change control. Apps edited live in a browser can break without warning. Appsmith offers Git integration on its free plan, and other platforms offer version history at various tiers. Our article on why no-code automations still need version control explains why the same discipline applies here.

Ownership. Name an owner for every app. When its creator leaves the company, an orphaned tool with database access becomes both a risk and a mystery.

Common mistakes

1. Treating the first app as a one-off. Internal tools accumulate. Choose a platform you can live with for years.
2. Choosing from demos. Build one real app against real data before committing. Demos hide friction in authentication, data modeling, and permissions.
3. Underestimating integration work. The canvas is the quick part. Getting a clean, secure connection to messy data takes longer.
4. A single maintainer. If one engineer holds all the knowledge, vacations and departures become outages.
5. Skipping the business case. If you cannot say how much time an app saves, it is hard to defend the cost or decide what to build next. Our automation ROI calculator gives a simple way to estimate it.
6. Automating a process nobody mapped. Write down the steps first. See how to map a business process before automating it.

How to pick: decision rules

• You want a working app soon and have budget: Retool. The free plan covers up to five users, which is enough to test a real use case.
• You prefer open source with a permissive license: Appsmith, under Apache-2.0.
• You want free self-hosting with no user cap and accept GPL v3: Budibase.
• You want a modern open-source option and can accept AGPL-3.0 terms: ToolJet.
• You need internal pages that share a React design system: Plasmic, with developers on staff.
• Your needs are mostly Python data apps: Consider Streamlit; see our data science tools roundup.
• Your needs are mostly scheduled jobs, not screens: A workflow tool from our best automation tools 2026 comparison may be enough, and combining internal tools with agents is covered in our AI agents vs traditional automation guide.

Frequently Asked Questions

Q: Is internal tooling really a separate category from no-code? A: Largely yes. Internal tools assume a database behind every screen, role-based access, and actions that write data. General no-code builders are more often aimed at public-facing apps.

Q: How does this compare to building with a framework like Next.js? A: For a few simple admin screens, a builder is usually faster. For an application that will change constantly for years, custom code gives more control and avoids platform limits. The right moment to switch is when you spend more effort working around the platform than using it.

Q: Are there free options worth using? A: Yes. Retool's Free plan covers up to five users, ToolJet's free plan allows two builders and up to 50 end users, Appsmith's Community edition is free to self-host, and Budibase's open-source edition has no cost beyond hosting. Each free tier has limits, so test against your real workload.

Q: What does an open-source license mean for me? A: Apache-2.0 (Appsmith) is permissive. GPL v3 (Budibase) and AGPL-3.0 (ToolJet) are copyleft; AGPL extends its obligations to software used over a network. Running the software unmodified for staff is a different situation from modifying and offering it to others. Ask a lawyer if you plan the latter. This is general information, not legal advice.

Q: Can these handle a high-traffic public app? A: They are designed for staff-scale usage. Retool prices external users separately on its Business plan, but a large public product is generally better served by custom software.

Q: Do they connect to Postgres and MongoDB? A: All five commonly connect to mainstream databases and REST APIs. The harder part is network security: VPNs, tunnels, and allowlists so the platform can reach the database without exposing it.

Q: What should a non-technical operations lead try first? A: Pick a small, real problem, such as a lookup-and-update screen for a spreadsheet the team already uses, and build it on a free tier. Writing down the process first makes the build much shorter.

---

Read next:

• Integrate Tools Without Coding — No-Code Guide
• Accounting Automation 2026 — AP, AR, Bookkeeping, and Close
• Workflow Automation Examples — Real-World Use Cases