Pricing
Priced on what you need validated.
Four surfaces, each with its own natural unit. Buy one on its own, or bundle them into a tier. Connectors are included plumbing, never a meter.
This page explains how RedMaw is priced. Scope drives the number, so the number comes from a conversation about your environment.
The model
Four surfaces. Four units.
Each surface is metered on the thing you already count, rather than on hosts.
Applications
Per application
External exposure, web and API testing, authentication and authorization, business logic, and secret exposure on the applications you have verified you own.
Why this unit
An application is the unit you already think in, and the unit whose blast radius you can reason about. Per-host pricing would punish you for how the application happens to be deployed.
AI Systems
Per model
Adversarial testing of a model, endpoint or AI feature registered as a first-class target: prompt injection, disclosure, jailbreaks, control bypass, posture, and release-gate comparison after model change.
Why this unit
Models are versioned, retrained and replaced on their own cadence. A single AI feature inside one application is covered by that application; the model meter is for systems you want tested as targets in their own right.
SaaS & Identity
Per identity
The customer-controlled identity and permission graph: roles, OAuth grants, tokens, service accounts, sharing, and reachable SaaS posture.
Why this unit
Your breach path is a compromised person reaching data through the applications they can access. Identity count is your real attack surface here, and a number you already know. You are never billed per SaaS vendor.
Internal Infrastructure
Per device
Authorized internal reachability from a foothold: reachable services, credential exposure and privilege boundaries, via the secure outbound connector.
Why this unit
A device is a distinct testable host. Each virtual machine or instance has its own operating system, credentials and foothold, which is what actually gets attacked. Containers roll up to their node.
Connectors are plumbing, not the product.
An integration is the infrastructure RedMaw needs to do its job. Metering it would charge you for connecting the systems that make the testing useful.
You buy security outcomes and scope. You should not have to count integrations.
Never metered
- Connectors and integrations: Jira, GitHub, Slack, identity providers, SaaS posture, DNS and CI
- The number of tests, runs or scans within your agreed scope
- Users and seats inside your workspace
- Findings, evidence or reports produced
- Retests after remediation
Tiers
Separated by capability, not by allowance.
Each tier changes what RedMaw tests and who operates it. Scope is agreed separately, per surface.
Core
Teams putting continuous adversarial validation in place for the first time, usually with security owned by engineering or by one person.
Capability
- External attack-surface discovery
- Web application and API testing
- Secret and credential exposure
- AI red-teaming against deployed models and endpoints
- Identity attack paths
- Validated exploitation with reproducible evidence
- Findings state, built-in tracker and auto-retest
- Continuous scheduling and deployment triggers
- Jira, GitHub Issues, Slack and email
- SSO, RBAC and audit logging
- Signed reports where supported
Operating model
Self-serve. Your team owns the operating loop.
Advanced
Most commonOrganizations with a real security function, or under regulatory pressure, that need evidence across more of the environment.
Capability
- Everything in Core
- Authorized internal testing via the secure outbound connector
- SaaS posture across customer-controlled tenants
- AI red-teaming in the CI or model-release pipeline, with baseline comparison
- Session recording and replay
- MITRE ATT&CK mapping
- Compliance evidence: PCI DSS 4.0, GDPR, SOC 2 and ISO-oriented programs, NIS2, DORA, EU AI Act
- Approval gates for higher-impact actions
- Open webhook and API
- Priority support
Operating model
Self-serve or managed. RedMaw operators can run and triage the program with you.
Enterprise
Larger and regulated environments where isolation, regional control or custom integration is part of the buying decision.
Capability
- Everything in Advanced
- Managed dedicated hosted deployment
- Regional hosting and customer-controlled key options
- Secure outbound connectivity for authorized internal reach
- SAML and SCIM
- Custom modules and integrations
- Deeper SaaS posture coverage
- Named operator engagement
Operating model
Managed or dedicated. Architecture follows the approved environment.
Roadmap: not shipped
- Detection-gap scoring and SIEM / EDR integration
- Attack-vs-detection timeline
- Automated cross-surface attack-path chaining
- Internal network agent and automated Active Directory attack-path discovery
- Red-teaming your AI agents, and deeper tool-abuse validation
- Source-code analysis
- OT and IoT testing
- In-tenant sovereign deployment
Commercial terms
How the agreement works
- Annual
- Scope is agreed for the term. Unlimited continuous testing within that scope, with no per-scan billing and no surprise metering.
- À la carte or bundled
- Each surface can be bought on its own, or bundled into a tier. Bundling is cheaper than the sum of the parts, so growth pulls toward the next tier rather than stacking add-ons.
- Scope you can audit
- Devices are discovered and counted by RedMaw at testable-host level, in-scope only, rather than self-declared, so the number is verifiable and you can see the exact list.
- Scope changes are transparent
- Applications, models and identity scope are add-ons with predictable terms. Tell us when the environment grows.
This is adversarial validation, priced against manual penetration testing rather than against per-asset scanning.
A scanner runs a library of known checks deterministically, and that coverage is worth keeping. RedMaw attempts exploitation within authorized scope, proves what is reachable, preserves the evidence and re-attacks the fix, across applications, identities and AI systems a host scanner was never designed to see.
Different job, different category, different price. Keep the scanner for coverage.
Questions
What buyers ask about the model
- Why per identity rather than per host?
- Because the breach path is a compromised person reaching data through the applications they can access. Identity count is your real attack surface on that surface, and it is a number you already know and can budget. Per-host pricing punishes large estates and undersells the value, since a real attack path spans many hosts. You are never billed per SaaS vendor.
- We do not own our SaaS platforms. Why pay to test them?
- We do not test the vendor's platform. That is theirs, and their own certifications cover it. We test your half of the shared-responsibility boundary: identities, permissions, OAuth grants, sharing and leaked keys. That is where breaches involving SaaS actually happen, and it is the part you are accountable for.
- How is AI priced?
- Per model. Each standalone model, endpoint or agent you want red-teamed is a unit. A single AI feature baked into one application is covered by that application. The model meter is for AI systems tested as first-class targets, including release-gate comparison when the model changes.
- How do you count devices?
- By testable host rather than physical box. A server running several virtual machines is several devices, because each has its own operating system, credentials and foothold, and that is what gets attacked. Containers roll up to their node; switches and routers count individually. You do not count them: RedMaw discovers and counts what is in scope, so the figure is auditable.
- Can we buy just one part?
- Yes. The four surfaces are available individually, so a SaaS company can take applications, an AI-native company can take models, and an infrastructure-heavy organization can take internal devices. Bundling into a tier costs less than buying the same scope à la carte.
- What happens if our scope grows mid-term?
- Applications, models and identity scope are transparent add-ons. The annual commitment covers unlimited continuous testing within the agreed scope, so ordinary growth does not produce a surprise bill. It produces a conversation about scope.
- Why is this priced above a vulnerability scanner?
- Because it is a different job. A scanner runs a library of known checks deterministically, and that broad coverage is worth keeping. RedMaw attempts to exploit within authorized scope, proves what is reachable, preserves the evidence, and re-attacks the fix. It also covers identity, SaaS and AI a host scanner was never designed to see. This is adversarial validation, priced against manual penetration testing rather than against per-asset scanning.
- Do we still need our annual pentest?
- Possibly, yes. For especially novel work, or where a customer, auditor or contract specifically requires a human-led third-party assessment, human testing remains the right answer. RedMaw covers the period between those engagements, when the environment actually changes.
Why there are no figures on this page
Because a published number would be a commitment we are not ready to make, and you deserve to know that rather than read a rate card we would revise.
Scope drives price, and scope varies more than a table can express. A company with one application and many identities and a company with many devices and one application are different engagements at similar cost.
Tell us what you need validated and how you want it operated, and you will get a number for your environment rather than a starting point you have to negotiate away from.
Tell us what you need validated.
Scope the applications, identities, models and environments that matter, and we will price the engagement around them.