Retool or Appsmith? The Real Cost of an Internal Tool

Do not commission a custom application just because your team needs an approvals screen. If the job is to read existing records, update a few fields and route work between colleagues, a platform may get you there with less development effort. The harder question is whether its subscription and operating costs still make sense once the right people and permissions are included.

A cheap plan that cannot separate a support agent from a finance administrator is not the plan to price for a refund tool. Equally, an expensive custom build does not become good value merely because it has no per-seat licence.

This guide compares a 25-person internal team, then shows how to put platform fees and custom development into the same budget. It complements our build-or-buy guide for web applications: the specific decision here is about an internal interface over your business systems, rather than replacing those systems wholesale.

What counts as an internal tool here?

Think of a staff-only order lookup, a supplier approval queue or a screen joining information from your CRM and billing system. The people using it are known colleagues, and the interface serves a defined task.

Start by distinguishing the screen from the rules underneath it. Displaying an order is straightforward. Deciding who may refund it, preventing duplicate refunds and recording why it happened are separate requirements. A visual builder can help with the interface; someone still has to design and verify those rules.

For a first version, write down one complete workflow. Who starts it? Which data does it read? Who approves a change? What happens when the connected system is unavailable? If those answers are missing, asking a developer to choose between Retool, Appsmith and custom code is premature.

You may also discover that the existing CRM already provides the required view. Configure that first if it solves the problem. Buying a second platform to recreate a feature you already pay for adds another system to maintain.

What do the published seat prices actually buy?

The figures below were checked on 3 November 2026. Dollar figures are the published dollar-denominated list prices, before taxes, conversion charges, discounts and additional usage. Confirm currency, region and billing terms in your actual order.

Platform and plan	Published monthly seat rate	Basis used below
Retool Team	$10 per builder; $5 per internal user	Annual billing
Retool Business	$50 per builder; $15 per internal user	Annual billing
Appsmith Business	$15 per user	Monthly rate

Sources: Retool pricing and Appsmith pricing. Retool lists audit logs and richer permissions on Business, while self-hosting and Git source control are listed under Enterprise. Appsmith Business lists custom roles, audit logs and unlimited Git repositories; its free plan lists three Git repositories and three standard roles. These are different packages, not equivalent seats with different prices.

Appsmith's pricing FAQ counts workspace members present at any point in the billing cycle for the full month, without an extra developer-seat rate. For Retool, confirm the billed builder/internal-user split for your organisation rather than equating every person with the cheapest seat.

Airtable is another possible route if the workflow fits its data model. Its pricing FAQ lists Team at $20 and Business at $45 per user per month on annual billing, with edit permissions determining billable users. Twenty-five staff do not necessarily mean twenty-five paid editors. Count the people who need the relevant permissions before comparing any platform.

What does a 25-person team pay in platform fees?

Assume three people are billed as builders and twenty-two as internal users in Retool. For Appsmith, assume twenty-five billed users. Hold those counts and list prices constant throughout the comparison; this is arithmetic, not a forecast of future vendor pricing.

Option	Monthly equivalent	One year	Three years
Retool Team	$140	$1,680	$5,040
Retool Business	$480	$5,760	$17,280
Appsmith Business	$375	$4,500	$13,500

For example, the first row is three times $10 plus twenty-two times $5. Annual-billing rows show a monthly equivalent, not a promise that you can cancel after a month. Read the commitment and renewal terms before signing.

The table excludes implementation, database hosting, connected services and the people maintaining the application. It also excludes additional AI or workflow usage. Calling these totals the cost of the finished tool would understate the budget.

If several tools serve the same team, check how the vendor counts users across them. Do not multiply a user-based bill by the number of screens without checking the licensing terms. Equally, include contractors, temporary staff and additional environments where your actual contract makes them billable.

Why do permissions matter more than a cheap seat?

Describe access using real tasks: support can read an order, finance can approve a refund, and only an administrator can change the connection settings. Test with separate accounts representing those roles.

Retool's permissions documentation distinguishes fixed permissions on Free and Team from configurable groups and object access on Business, with further controls on Enterprise. That is why the lowest price is not a valid baseline until the access model fits.

Appsmith's granular access-control documentation describes permissions on resources such as applications, queries and data sources, bundled into roles. Map those controls to your requirements; do not assume that being able to assign an app role automatically implements every business rule about individual records.

Use a negative acceptance test as well as a successful one. Can a support user approve the refund by a different route? Can someone access another department's records? Does a broad database credential bypass a restriction that exists only in the interface? The checks belong in the project scope whichever platform you choose.

An audit feature needs inspection too. Appsmith's audit-log reference documents events and their associated user, time and resource information. Ask whether your specific approval decision, reason and record change are captured. A platform log is useful evidence, but its existence does not prove that your complete business history has been recorded.

How do you compare that with a custom build fairly?

Use the same time horizon and include labour on both sides. For a platform, count implementation, subscriptions, application maintenance and connected infrastructure. For custom development, count the build, hosting, monitoring, upgrades and future changes.

Here is a fictional budget, not a quotation or a claim about market rates. Assume ₹90 per dollar solely to make the arithmetic visible. Suppose platform setup costs ₹60,000 and ongoing application upkeep costs ₹36,000 per year, in addition to the fees above. Suppose a comparable custom application costs ₹6,00,000 to build and ₹1,20,000 per year to operate and maintain from year one. All amounts exclude tax.

Three-year budget at those assumptions	Total
Retool Team fees + setup + upkeep	₹6,21,600
Retool Business fees + setup + upkeep	₹17,23,200
Appsmith Business fees + setup + upkeep	₹13,83,000
Custom build + three years of operation	₹9,60,000

For the first row, $5,040 times ₹90 is ₹4,53,600; adding ₹60,000 and three years of ₹36,000 produces ₹6,21,600. The custom row is ₹6,00,000 plus three times ₹1,20,000.

In this example Team is cheaper than custom, while the two higher-cost options are dearer. But if Team fails the permission requirements, its attractive total is irrelevant. And if the custom quote excludes equivalent permissions, logs or recovery work, that comparison is also invalid.

Replace every assumption with your own quote and measured workload. Our custom CRM pricing page covers a related project category, but a CRM starting price is not a quotation for your particular internal tool. There is no universal seat count at which custom becomes cheaper.

Does self-hosting remove the subscription problem?

It can remove some licence costs, but it creates operating responsibilities. Appsmith lists a free open-source Community edition. Confirm the features and deployment terms of the edition you intend to run, rather than transferring paid-plan assumptions to it.

Budget for a named person to install updates, restrict network access, manage credentials, monitor failures and restore the service. Also identify which connected databases need their own backup. Recovering the visual application does not necessarily recover the business records it edits.

Ask for a recovery demonstration during the pilot. If the only person who understands the installation is unavailable, can another maintainer restore it from the documentation? A free licence with an unowned recovery process is a weak foundation for an important daily workflow.

Choose self-hosting because its control and operating model suit the business, not because the licence line happens to be zero. If your team cannot support it, include managed operation in the comparison or choose a hosted service.

Can a low-code application be reviewed and maintained properly?

Yes, if you deliberately set up the process. It is inaccurate to dismiss every visual platform as unreviewable. Appsmith's Git documentation describes versioning, branching and collaboration. Plan availability still needs checking, but low-code does not mean abandoning change control.

Before accepting a tool, ask the maintainer to demonstrate a small change, review it, release it to a test environment and recover the previous version. Separately test what happens if the change also modifies the underlying database: rolling back the interface may not reverse a data migration.

Keep business rules in a maintainable place. A long expression hidden inside a button can be difficult to inspect; the same would be true of poorly structured custom code. Require clear ownership, a short architecture note and tests for the operations that could damage records or move money.

For custom web development, those responsibilities do not disappear. The business case must include someone maintaining the application after launch, rather than treating delivery as the end of the spending.

When should you avoid either option?

Avoid a custom build when the process is still changing and a small prototype would answer the important questions. Avoid a platform commitment when a required integration, permission boundary or export path has not been demonstrated.

Do not rule out platforms simply because some users are external. Evaluate the permitted deployment, access controls, branding and actual external-user pricing. Customer-facing use is a different requirement to price and test, not an automatic disqualification.

Do not put live customer records or payment access into an early prototype just to make the demonstration convincing. Representative test data can establish whether the workflow fits before production permissions are granted.

Finally, do not build another interface around a process that should be removed. If staff repeatedly copy data because two existing tools are not connected, a focused integration may solve the problem without a new daily application.

What should you do before requesting a build quote?

Pick one workflow and list its users, editors, approvals and failure cases. Prototype it with representative data. Test the required permissions and recovery process. Then compare a written platform implementation proposal with a custom proposal covering the same behaviour over three years.

Choose the option that meets the requirements at an acceptable operating cost. That may be a platform, a small integration or a custom application; the budget should explain the decision rather than decorate a preference you already had.

Tell us the workflow and who needs access if you want a second opinion. If a modest platform setup is sufficient, there is no reason to sell you a larger build.