Retool Enterprise: Security Controls Buyers Should Verify
Enterprise app governance is not a checkbox. Learn what to verify around identity, permissions, auditability, and the software layer where agent work actually runs.

Retool Enterprise is Retool's tier for organizations that want internal apps with centrally managed identity, permissions, and audit records. Before you expand it past a pilot, verify six things in your own deployment: SSO and group mapping, app and resource permissions, credentials, audit-log access, and what happens when a role changes or someone leaves. Test them with real accounts, because a feature name on a pricing page tells you what exists, not what your auditor will be able to reconstruct.
Major takes a different starting point. On Major, SSO, role-based access, scoped credentials, and audit logs ship with every app and agent at the platform layer, so each new workflow inherits security instead of rebuilding it. The rest of this piece covers what to test in any internal-app environment, why the test belongs at the point of action, and how to build a reviewable workflow on Major.
What enterprise buyers need to verify in Retool
Start with what Retool itself says. Its enterprise page describes identity-provider sign-in with automated role mapping, audit records of user activity, role-based and data-level access restrictions, cloud or self-managed deployment, and secret rotation through a Secrets Manager and API. The permissions documentation says access to apps, resources, workflows, and agents is controlled through groups, with use, edit, and own levels, and that a user's access combines the permissions of every group they belong to. It also states that custom groups and some finer controls depend on plan, with Enterprise adding finer-grained admin roles and broader organization-level settings. The SSO and user administration docs cover enforcement, group sync, and user management.
Those are vendor statements about capability. Your deployment is what matters, and controls and availability depend on configuration and plan. We have not audited Retool, and nothing here says a control is missing.
Two individual reports show the kind of worry buyers bring. One Reddit user reports concern about "bad permissions" and "random AI-generated architecture" when comparing tools. One Medium author reports that public apps require a Business or Enterprise cloud plan and calls those plans expensive. Neither is evidence of a platform-wide defect. They do show why buyers should check plan gates against their own use cases before committing.
Scope note: we did not pull a live SERP for this piece, and no compliance certification is implied for Major.
A practical governance check for internal apps and agents
Run this against any internal-app environment, Retool included. Use representative accounts, not an admin login.
- Identity and SSO. Is SSO enforced, and can anyone still sign in with a local password?
- Group and role mapping. Do identity-provider groups map to platform groups automatically, and does a membership change in the IdP show up in the platform without manual cleanup?
- App, resource, and workflow permissions. Can a user with "use" access reach a data source that the app queries on their behalf, and can they edit it?
- Credentials. Whose credentials does an app use at query time, how are they scoped, and who can read them?
- Audit-log access and review. Can a security reviewer retrieve who did what, when, and under which role, without an admin ticket?
- Role change and offboarding. Remove a user from a group and disable the account. What access remains in shared apps, scheduled jobs, and API tokens?
Then test one least-privilege workflow end to end. This three-row example shows the shape of the evidence:
- Builder: can create and edit the exception-review app. Evidence to review: change history and who published each version.
- Operator: can view and work the queue, cannot change thresholds. Evidence to review: the actions taken and the account each ran under.
- Approver: can approve or reject sensitive actions. Evidence to review: decision, actor, and timestamp on every approval.
A platform control, such as SSO or an audit log, applies everywhere without anyone remembering to add it. An app-specific process rule, such as "finance exceptions over a threshold need a manager," exists only because someone built it into that one app. Buyers often verify the first and assume the second.
Why controls must follow the work
Enterprise readiness shows up where an app or agent acts. A control list can be complete and an action path can still be opaque, if the logic that decides what happens lives in a prompt or in one-off glue code that nobody can inspect.
The NIST AI Risk Management Framework organizes risk work around govern, map, measure, and manage, and NIST SP 800-53 catalogs access control and audit controls. Neither document tells you which product to buy. Both point at the same operational question: can you attribute an action to an actor and reconstruct it later? That gets easier when repeatable work runs as deterministic code that holds its own state and writes its own records, and harder when each run re-reasons the task from scratch. Our pieces on governance for agentic AI, scoping AI agent access, and agent observability go deeper on each half.
How to build this on Major
Here is a proposed pattern for the finance-exception scenario. It is illustrative, not a live customer deployment: an employee requests an exception, an agent drafts a recommendation, a manager approves, and the app records the decision and the actor.
- Connectors. Salesforce is the system of record for the account and contract. Slack carries manager approvals and alerts. PostgreSQL holds the app's review queue and decision history. Each connector uses its own scoped credential, so the workflow reaches only what it needs.
- The app holds state. A review app on Major's managed database stores each exception, its status, the threshold applied, and every decision. State lives in the app, not in a conversation that ends.
- Deterministic logic. App code applies the thresholds, routes the case, and records status. This part runs the same way every time and uses no model.
- The agent's judgment. The agent only classifies or summarizes ambiguous cases, such as a request that does not fit a standard category. It drafts a recommendation; it does not approve anything.
- Trigger. A scheduled workflow checks pending exceptions and posts any that need a decision to the manager in Slack.
- Governance. Role-based access separates reviewer from operator. SSO covers every sign-in. Audit logs record who approved what and which credential was used.
This is the same reason the agent reasons once and the app runs from then on: the repeatable steps become software with permissions and logs attached. You can start building free and wire up this pattern with your own connectors.
What we're doing about this at Major
Security should travel with the app and the agent. It should not be rebuilt for every generated workflow, because that is where gaps appear: one app gets a role check, the next one forgets it. So Major ships every app and agent with SSO, role-based access, scoped credentials, and audit logs at the platform layer.
We hold a position here that you can disagree with. A governance review should start from how a workflow acts, with a named actor and a retrievable record. It should not start from a feature matrix. Major does not replace your identity lifecycle, your review procedures, or your compliance program, and its listed controls are not a guarantee of compliance. They make those processes easier to run because the evidence already exists.
Start building on Major
Build a governed internal app that keeps exception state and review history in one place, with an agent handling judgment and a person approving sensitive actions.
Related articles
Frequently asked questions
- what is an enterprise tool?
- An enterprise tool is software used across an organization that comes with administration, access controls, and operational oversight. For internal apps, that means SSO, role-based permissions, credential management, and audit logs you can retrieve. Buyers should test those controls with real roles before expanding adoption.
- how big is the retool company?
- Company size changes over time, and the sources behind this article do not establish a current figure, so we do not state one. For current headcount or funding details, check Retool's own company pages or recent official announcements rather than relying on a secondhand number.
- what is the most popular enterprise software?
- Popularity depends on the category, region, and metric, so no single product holds a universal ranking. For internal apps and agent-driven work, a better question is which platform gives you SSO, role-based access, scoped credentials, and retrievable audit evidence for every workflow you run.