Design-partner applications

Human-validated attack-path assurance for SaaS and AI products

Start with a specialist-led baseline assessment. Then keep important roles, tenant boundaries, APIs, AI tools and previously tested controls visible as the product changes. Selected design partners can help shape a shared workflow for scope, evidence, remediation and re-testing.

Clear availability:

This is not autonomous production testing or a self-service scanning platform. Baseline assessments and scheduled testing are available by written scope. A shared validation workflow is in development for selected design partners.

Offer status

What you can engage now - and what is still being built

Availability labels keep an assessment service, a scheduled cadence and a software workflow from being mistaken for the same thing.

Available by scope

Baseline assessment

Web, API, SaaS and AI application testing with a written boundary, human-verified findings and agreed deliverables.

By written scope

Scheduled validation cycles

Selected change reviews, regression of important controls and agreed re-tests after the baseline.

Design partner

Shared validation workflow

A planned service-backed workflow connecting scope, evidence, finding status, ownership and re-test decisions.

Planned

Repository-assisted change review

A possible later capability. It is not currently offered as automated pull-request review or a public product.

Start with evidence

Recurring penetration testing starts with a reliable baseline

A recurring plan is only useful when everyone knows what the earlier assessment covered. The baseline records the agreed application boundary, relevant roles and tenants, critical workflows, APIs, integrations, environment, exclusions and tested version.

After that, each cycle answers a smaller question: what changed, which attack paths deserve another look, which fixes should be verified, and what remains outside the current boundary?

Compare the delivery options on the penetration testing as a service page, or draft a private web or API scope.

The validation model

Attack-path validation across identity, tenants, APIs and AI tools

The work follows the boundaries an attacker would need to cross, not a generic list of checks. Every cycle begins and ends with an explicit decision.

  1. 01 / Define

    Set the boundary

    Record assets, roles, tenants, critical workflows, test identities, authorization and limits.

  2. 02 / Validate

    Prove the path

    Test selected attack paths and manually reproduce issues before they reach the report.

  3. 03 / Remediate

    Keep context attached

    Connect evidence, severity reasoning, ownership and practical remediation to each finding.

  4. 04 / Re-test

    Check the change

    Verify agreed fixes and controls against the version actually tested, then record the result.

Accountable delivery

A human review gate before findings reach your team

Automation may support research, repeatable checks and evidence organization. A researcher remains responsible for scope, safe execution, impact validation, severity and delivery.

Candidate observations do not become reported findings until they have been reproduced, checked against scope, reviewed for impact and prepared with evidence your engineers can follow.

Inspect the structure in the fictional sample penetration testing report and read the AI-use and human-accountability approach.

One connected record

Evidence that stays connected to remediation and re-testing

The goal is not another dashboard. It is a usable chain from the approved boundary to a confirmed issue, the fix decision and the result of checking that fix.

SCOPE

Versioned boundary

Assets, roles, tenants, workflows, accounts, exclusions and test limits for the current cycle.

PROOF

Reproducible evidence

Steps, requests, responses and impact notes with secrets and unrelated data redacted.

DECISION

Remediation ownership

Severity reasoning, recommended action, owner, status and open questions kept with the finding.

VERIFY

Re-test history

The version checked, evidence from the re-test and a clear fixed, partial, open or not-tested result.

Bounded by design

Safety controls for every testing cycle

Recurring work does not create standing permission to test anything at any time. Each cycle has a written boundary and stop conditions.

Required before testing

  • Versioned target and environment list
  • Controlled identities and synthetic test data where practical
  • Approved actions, request limits and testing window
  • Emergency contact, stop conditions and escalation path
  • Data handling, retention and AI-use decisions

Excluded by default

  • Denial of service and destructive actions
  • Persistence or lateral movement outside the agreed boundary
  • Social engineering and physical testing
  • Access to real customer data beyond minimum safe proof
  • Testing third-party assets without their authorization

Read the full scope, evidence and client-data standards.

Private design-partner program

Help shape a more useful validation workflow

The pilot is intended for B2B SaaS or AI product teams with multiple roles, tenant boundaries, APIs or agent tools, and a need to validate important changes more often than an annual assessment.

Tell us what changes frequently, which attack paths matter most and what evidence your engineers or reviewers need. We will first confirm whether a baseline assessment or a design-partner conversation is the right next step.

Recurring security validation FAQ

Is this an autonomous penetration testing platform?

No. Current delivery is specialist-led. Automation may support approved research, repeatable checks and evidence organization, but it does not expand scope or replace human validation.

What is available now?

Baseline web, API, SaaS and AI application assessments are available by scope. Scheduled change cycles and re-tests can follow a baseline when capacity, access and authorization are agreed.

What is the design-partner program?

Selected teams can help shape a planned shared workflow that keeps scope, evidence, finding status, remediation ownership and re-test decisions connected. This workflow is in development; availability and features are confirmed during the design-partner conversation.

Does recurring validation mean continuous or 24-hour testing?

No. Testing happens in agreed windows against a written boundary. The cadence, response expectations and work reserved for each cycle are defined in the engagement agreement.

Choose the honest starting point

Baseline assessment or design-partner conversation?

Share the system, roles, tenant model, release cadence and the evidence you need. We will confirm the next useful step before proposing work.