Services

Three steps to a copy of your systems we can attack.

The Survey plans a sandboxed copy of your production systems. Red Team Setup builds it and attacks it with a swarm of AI agents. Standing Red Team keeps attacking it as your systems and attackers’ models change.

Request a free scoping call

The Survey Red Team Setup Standing Red Team

After the scoping call · Before any quote

The Survey

A forward-deployed engineer works alongside your team, remotely, to understand your systems, then writes the action plan for your mock environment. We agree the survey’s terms on the scoping call.

What happens

  1. Kickoff. Your named contact walks us through your architecture and who owns what. We agree a cap on your team’s time.
  2. We read before we ask. If your code or docs answer it, we don’t ask. Questions come batched, one message a day.
  3. Action plan. We write the plan for your mock environment. Your fixed quote for Red Team Setup follows.

What you get

  • A written action plan: how we build your mock environment, the easiest way for your team to help, and where it runs
  • The outside services you depend on (login, payments, email, records) that need a test account or stand-in
  • A list of what’s running in your cloud that isn’t in your code

What we need from your team

  • A chat channel with your engineers, ideally a guest account in your own Slack or Teams
  • One named contact and one kickoff call
  • Docs you already have: architecture diagrams, API specs, your last pentest report
  • Read-only access to your infrastructure code
  • View-only cloud access, recommended: production drifts from the code, and that gap is where problems usually are
  • Source code: recommended, not required. With it, our agents find more

Never patient data, write access to your systems or production credentials. How we handle access

One-time · Fixed quote after the survey

Red Team Setup

We build your mock environment from the action plan, attack it with our agent swarm and give you a verified findings report, with one re-test included.

What happens

  1. Build. We build the mock environment: a sealed, sandboxed copy of your production systems, as realistic as possible, with synthetic data.
  2. Attack. Our AI agents attack it autonomously, at full aggression.
  3. Verify and report. An engineer reproduces every finding before it goes in your report.
  4. Re-test. After you fix the findings, we run the red team once more to confirm the fixes hold.

Who it’s for

MedTech companies that want to know how their backends, APIs, cloud and web apps stand up to AI-driven attacks, without risking production.

What you get

  • A findings report: prioritised vulnerabilities with attack paths, reproduction steps and fix guidance
  • No false positives: every finding verified by an engineer
  • One re-test to confirm your fixes hold

What we need from your team

  • A completed survey: we build from its action plan
  • Help standing it up, following the plan
  • Fixes for the findings, before the re-test
Ongoing · Monthly fee · After Red Team Setup

Standing Red Team

We keep your mock environment matching production and attack it again every month, and whenever a significant new open-source model is released.

What happens

  1. We keep the copy current. As production changes, we update the mock environment to match.
  2. Monthly run. The agents attack the environment once a month.
  3. New-model runs. When a significant new open-source model is released, we run the red team again.
  4. Updated report. Every run produces an updated report, every finding verified by an engineer.

Who it’s for

Teams that have done Red Team Setup and want to keep up as their systems change and attackers get better models.

What you get

  • An updated, verified findings report after every run
  • A mock environment kept in step with production
  • A red team run against each significant new open model

What we need from your team

  • A completed Red Team Setup: we reuse its environment
  • A way for us to keep up with changes to production, agreed with you

What we attack

In scope

  • Backend services and APIs
  • Cloud infrastructure and data stores
  • Web apps and portals, attacked through a real browser driven by our agents

Not in scope

  • The devices themselves: on-device software and physical or wireless interfaces
  • Mobile apps
  • Supply chain
  • Social engineering

Why a real browser? Some attacks only exist inside a browser: cross-site scripting, cross-site request forgery and broken login or session flows.

What the copy leaves out. Web firewalls, CDNs and bot filtering usually aren’t copied, so our test is harsher than a real attack. Findings your firewall may block today are labelled, and still worth fixing.

How it runs

Where it runs

A throwaway cloud account you own, deleted afterwards, or hosted by us. The action plan settles it.

Your data

Synthetic data and fresh secrets. No real patient data. De-identified data only if you want it, agreed case by case.

Sealed off

No internet. Agents reach a hosted AI model through one logged exit; we name the provider before any run. How we seal it

What you keep

The action plan, every report, and a mock environment we keep set up and operate for later runs.

How the three fit together

  1. The Survey comes first. You get the written action plan. We agree its terms on the scoping call.
  2. Red Team Setup follows. It builds your mock environment and gives you your first verified report and one re-test. A fixed quote, given after the survey.
  3. Standing Red Team, optional. It keeps the same environment matching production and attacks it every month and after each significant new open model. A monthly fee.

No red team run finds everything. What we can tell you is what our agents found, and that each finding is real.

Start with a free scoping call

Tell us what you run. We’ll tell you what it would take to copy it and attack it.

Request a free scoping call