Amazon Verified Permissions alternatives

AAlex OlivierNovember 16, 202610 min read
Amazon Verified Permissions alternatives

If you are looking for an Amazon Verified Permissions alternative, Verified Permissions is probably already doing a job for you. One application on AWS calls it, the policies work, and the bill for single checks is small. The question comes when authorization has to do more. A product team wants a list of everything a person can see. A second application needs the same rules. A business unit runs on another cloud, or an auditor asks which rule allowed a particular request last quarter. The alternative has to evaluate the same rules, and do the work Verified Permissions leaves to your code.

What Amazon Verified Permissions does and who uses it

Amazon Verified Permissions is AWS's managed authorization service, generally available since June 2023. It evaluates policies written in Cedar, an open source policy language AWS created, which Cerbos vs Cedar compares with Cerbos policy. You create a policy store, add a schema and Cedar policies, and your application calls the IsAuthorized API with the principal, the action, the resource and every entity the policies need. Verified Permissions returns allow or deny, and the policies that determined it.

It suits teams building custom applications on AWS, often with Amazon Cognito for sign-in and Amazon API Gateway in front, where AWS documents a Lambda authorizer that calls the service. AWS cut the price of single checks by up to 97% in June 2025, to $5 per million (AWS pricing page, October 2026), and added policy store aliases for multi-tenant applications in April 2026.

Why teams look for an Amazon Verified Permissions alternative

Lists of what a person can access. Verified Permissions decides one request at a time, or up to 30 in a batch call, and returns decisions rather than a filter. To show a person every record they can see, sorted and paginated, the application loads candidates and checks them in batches. ACES, a Tokyo AI company that compared Verified Permissions, OpenFGA and Cerbos, wrote (in Japanese) that the 30-item limit and the missing filter were the main reasons it ruled out Verified Permissions for its meeting product.

Every fact travels in the request. The application sends every principal, resource, parent and attribute the policies read, up to 1 MB per request (AWS documentation, October 2026), and Verified Permissions fetches nothing itself. A user's groups, a document's folders and a customer's account status all have to be loaded by every caller first. A Hacker News commenter wrote in 2023, "Cedar/AVP expects you to send all the contextual data as part of the Authz request". AWS also caps parents at 100 per entity, and that limit cannot be raised (AWS documentation, October 2026).

Decisions run only in AWS. Every check is a call to an AWS Region. The only ways to evaluate locally are experimental open source libraries. Teams with workloads on another cloud or on-premise need the same rules there.

Evidence for auditors. Decision logs are CloudTrail data events, AWS's per-request audit records, which are not logged by default, and a logged event records the decision without the policies behind it (AWS documentation, October 2026). Testing happens in a test bench in the AWS console, and versioning comes from the pipeline you build around it.

One store per tenant. For B2B products, AWS documents a shared policy store or one store per tenant, and with per-tenant stores every shared policy change has to be replicated to each store.

Who needs an Amazon Verified Permissions alternative, and how urgently

Not everyone needs to move now. A team running one AWS application, with a few rules, entity data that arrives with each record and no permission-filtered lists, can plan the move rather than rush it.

It is more urgent if any of these is true:

  • A product team needs lists or search filtered by permission.
  • Your code assembles deep hierarchies or large group memberships for every call.
  • Authorization has to run outside AWS, on another cloud, on-premise or at the edge.
  • An auditor, a regulator or your own security team asks for the policy behind past decisions.
  • You run a policy store per tenant and shared changes are getting harder to roll out.
  • Gateways, data platforms or AI agents need the same rules as your applications.

What to look for in an Amazon Verified Permissions alternative

Ask any replacement these questions.

  1. Where do decisions run? In one cloud's API, or next to each workload, on AWS and anywhere else?
  2. Who gathers the facts? Does every caller assemble the data, or can the platform fetch attributes from the systems that own them when the decision is made?
  3. Can it answer "what can this person see"? As a filter your database applies, or only as checks one item at a time?
  4. Is every decision on record? With the inputs, the outcome and the policy version, collected from every decision point and searchable months later?
  5. How is a change reviewed and tested? In Git, with tests that run in CI and block a release when they fail?
  6. How does it handle tenants? One set of policies with per-tenant exceptions, or a copy per customer?
  7. How far does one set of policies reach? Applications only, or also API gateways, data platforms and AI agents?
  8. Does it follow open standards? Does the decision API implement AuthZEN, so your enforcement points do not change if the decision service does?
  9. What does it cost at your real scale? Count the engineering time around each check, not only the check.

For a full breakdown of what to look for when selecting an authorization solution - refer to our detailed framework.

The main types of Amazon Verified Permissions alternative

Lists of Verified Permissions alternatives mix in identity providers and other AWS services. Only some of them evaluate the rules Verified Permissions evaluates.
![][image2]

Identity providers and AWS IAM. Amazon Cognito, Okta and Entra ID handle sign-in and put groups in a token. AWS IAM governs access to AWS resources such as S3 buckets and DynamoDB tables. Neither decides whether this person may approve this invoice in your application, so neither replaces Verified Permissions.

The Cedar library on your own. Cedar's open source libraries evaluate the same policies without the managed service. You then build and maintain the service around them, which means policy distribution, a testing pipeline, decision logs and the data assembly. Cedar alternatives covers what that takes, and the options beside it.

General policy engines and languages. Open source engines that evaluate policy written in their own language. They make fine-grained decisions, and on their own they leave distribution, testing, audit collection and data loading to your team.

Relationship-based access control (ReBAC) engines. Services modeled on Google's Zanzibar paper answer from relationships kept in their own store, such as which user is an editor of which folder. That is a second copy of your data to keep in sync.

Building your own. Authorization written into each application. It works until the number of services, tenants and auditors grows. Our post on build vs buy covers where that line falls.

Authorization management platforms. A decision engine plus the management around it, policy authoring and testing, distribution to every decision point, decision logs, and context from your other systems. This is the type that replaces Verified Permissions on AWS and beyond it, and it is where Cerbos sits.

Cerbos as an Amazon Verified Permissions alternative

Cerbos is an authorization management platform. Decision points run next to every workload, Cerbos Hub manages policy, testing and decision logs, hosted or on-premise, and Cerbos Synapse brings in context and connects gateways, data platforms and AI agents.

Runs on AWS and everywhere else

Cerbos decision points run inside AWS Lambda, on ECS, EKS and EC2, and on any other cloud or on-premise, so a decision is a local call next to the workload. Cerbos works with Amazon Cognito and Amazon API Gateway, and you can buy it through AWS Marketplace so the spend can count toward your AWS commitment.

Lists as a database filter, instead of 30-item batches

Query plans hand your database the policy as a condition it can apply, through adapters for the common ORMs, so lists come back sorted and paginated. ACES translated the same plans into Elasticsearch queries.

Facts fetched at decision time, instead of entity lists

The application sends what it has, and Cerbos Synapse fetches the rest from the system that owns it, including Cognito groups and attributes, directories, databases and internal APIs, with caching set per source. A team change in the directory applies once that source's cache refreshes, with no caller to update.

Decision logs linked to the policy behind them

A Cerbos decision log records the inputs and outcome of every decision, linked to the policy version behind it, and Cerbos Hub collects the logs from every decision point into one searchable, exportable trail. Retention is three months on Development, a year on Production and custom on Enterprise.

Policy in Git, tested in CI, instead of a console test bench

A Cerbos policy is short YAML with conditions in CEL, reviewed in pull requests. Test suites run in any CI system, and Cerbos Hub only builds a deployment when they pass. You can roll back to an earlier build in one step.

One set of policies across tenants

Scoped policies give a tenant its own exceptions on top of the shared rules, so a change to a shared rule ships once for every customer. A share is a row on the record, not a new policy.

One policy set for applications, gateways, data platforms and agents

The same policies cover applications through SDKs, Amazon API Gateway, Kong and Apigee through the Cerbos API or AuthZEN, Envoy and Istio through Cerbos Synapse, Trino row filters and column masks, Kafka topics and Kubernetes admission control. For AI agents, every tool call is checked down to its arguments, in the agent through hooks, at an MCP gateway or in the MCP server. AWS's own Cedar product for agents, Policy in Amazon Bedrock AgentCore, checks the calls that pass through its gateway, so agents outside it need their rules set somewhere else.

Open standards and published pricing

Requests to Cerbos can follow AuthZEN, the OpenID Foundation's standard Authorization API. Alex Olivier, Cerbos cofounder and CPO, co-chairs the OpenID AuthZEN Working Group, so read our AuthZEN claims with that in mind. Pricing is published, by monthly active principals, with no charge per decision.

Cerbos at enterprise scale

Customers send the Cerbos platform more than 2 billion authorization checks a month, and at Utility Warehouse it governs access across more than 4,500 services and thousands of non-human identities, with millions of access decisions a day. Cerbos holds a SOC 2 Type II attestation.

Analysts follow the category. Cerbos is recognized in the AuthZEN category of the Gartner® Hype Cycle™ for Digital Identity, 2026, and appears in KuppingerCole Leadership Compass reports on policy based access management, API security and management and generative AI defense, among others.

Moving off Amazon Verified Permissions without a big bang cutover

You do not have to move everything at once, or leave AWS.

  1. Export what the policy store holds. Verified Permissions has no export command, so list the policies, templates and schema through the API, one store at a time.
  2. Write the rules as Cerbos policies, with tests. Write each rule from its intent, and turn the decisions you know must hold into a test suite.
  3. Decide where each entity comes from. Keep sending what the application already loads, and let Cerbos Synapse fetch the rest.
  4. Run Cerbos in shadow mode. Ask both, keep enforcing Verified Permissions' answer, and log both until they match.
  5. Switch over one service at a time, then delete its policy store.

The migration guide walks through each step, from the export script to the IsAuthorized and BatchIsAuthorized replacements, and Cerbos vs Amazon Verified Permissions compares the two section by section.

For most teams whose authorization has grown past one AWS application, Cerbos is the right Amazon Verified Permissions alternative, because one set of rules runs on AWS and wherever else your workloads run, for every application, gateway and agent.

As Engin Attar, Head of Product and Growth and cofounder at Debite, put it, "[Cerbos] allows us to offer new products faster, because writing new policies or editing them is so easy."


Try Cerbos with the rules from one of your policy stores. Your engineers get permission-filtered lists from the database instead of batching checks, your identity team gets one set of rules for every tenant and every cloud, and your security team gets every decision logged with the policy version behind it.

To plan the move from Amazon Verified Permissions, book a call with our team, walk through the policy stores you run today, and leave knowing which service to move first and what your security team will be asked to sign off.

Go deeper:

FAQ

Tagged in

Free policy workshop

Get your first Cerbos policy written by our team.

Book a session to talk through your requirements and walk away with a working policy.

Book a session