Rupt vs. Castle
Castle offers similar functionality to Rupt but differs in the depth of the offering and how much you build on top of it, and how customizable each product is.
See Rupt on your traffic
Book a 20 minute demo. We walk through your use case on real evaluations and price out your volume.
The short version
Updated August 2026
- Castle focuses on the decisioning layer for well-defined workflows without the need for tailored customizations.
- It provides a good rules engine, and many of the important signals. It lacks the challenge engine, deeper signals, and offers no AI agent capabilities.
- Rupt is a full fraud engine. It includes fingerprinting, email and phone intelligence, a rules engine, hosted challenges, and an AI agent that investigates accounts and monitors for new patterns.
- Overage is $0.005 per call on both. The difference is how much of the stack is included in that number and what kind of support you get.
Capabilities.
Checked against both products' public pricing pages and docs, August 2026.
Why choose Rupt.
A full fraud engine, end to end
Identification, intelligence, decisioning, and enforcement are one product, not four you wire together. One evaluation returns the fingerprint, the signals, the scored risks, and a verdict, and Rupt acts on that verdict. Nothing in the middle is left for you to build and then maintain.
100ms at p99
Not an average that hides a long tail. 99 out of 100 evaluations come back inside 100ms, which is what lets you put Rupt directly in the login and signup path instead of running it after the fact.
Risks you can actually build
Rupt ships with its own risks ready to go, and you can compose your own from individual indicators with your own weights and your own actions. That is real customization, not tuning a threshold on somebody else's score.
Unmatched support
Integration setup help, developer meetings, and direct Slack and email access, with replies in hours, on every plan. Most vendors reserve that for their top tier. Don't take our word for it though, just ask our customers.
State of the art AI
The agent protects and detects. It watches your traffic for new and emerging fraud patterns as they form, then tells you what it found and what to do about it, down to the policy changes and updates it recommends you make.
A challenge engine that's a whole journey
Not a single screen bolted onto a verdict. Challenges are customizable and built around an end goal, whether that's adding friction, converting a sharer into their own account, or stopping a takeover, with cooldowns and velocity controls shaping how and when someone gets challenged.
Single-vendor pricing
Fingerprinting, email and phone intelligence, rules, and challenges usually mean a vendor and an invoice each. Because Rupt ships them together, the bundle costs less than the sum of the point tools.
Migrating from Castle.
- 01
Swap the client SDK. Load Rupt's snippet and call evaluate() on login, signup, and the actions you care about, the same places you generate Castle request tokens today.
- 02
Replace the risk call. Where your server sent a request token to Castle, read the Rupt evaluation instead: one GET returns the verdict, named risks, fingerprint, and device ID. Castle's numeric risk scores map to Rupt's named risks and allow / challenge / block verdicts, so your score thresholds become policy conditions.
- 03
Run both vendors for two to four weeks. Device IDs won't map one to one, so let returning users re-identify while you compare Rupt verdicts against Castle scores on the same traffic.
- 04
Turn on policies in observe mode, watch the verdicts, then enforce. If you built a step-up flow for Castle, you can retire it: a challenge verdict hands off to Rupt's hosted flow.
Frequently asked questions.
See Rupt on your traffic.
Book a demo and we'll walk through your use case, show you the signals on real evaluations, and price out your volume.
- A walkthrough tailored to your use case.
- How detection, the rules engine, and challenges fit your stack.
- Pricing and a rollout plan for your volume.