Seven beliefs that shape how we approach every medical device cybersecurity engagement. Each belief serves a value - and is enacted by one or more of our operating principles.
Every decision we make is filtered through the question: does this protect the patient using the device?
When commercial pressure, deadlines, or convenience collide with a real safety question, patient safety wins. We escalate hazards early even when it slows a submission or an engagement.
Scenarios from our work
Situation: A pen-tester finds an unauthenticated BLE command that can silence an infusion pump alarm two days before the client's FDA submission deadline.
What this looks like
Immediately notify the client's PM and cybersecurity lead, add the finding to the risk register with a clinical hazard analysis, and recommend a submission hold or targeted mitigation - even though it delays the client's launch.
Not this
Downgrading the severity so the finding fits into the residual risk table and the deadline stays intact.
Situation: A client asks us to remove a Critical CVE from the SBOM report because the affected component is 'not really exploitable'.
What this looks like
Keep the CVE, document the exploitability rationale under ANSI/AAMI SW96, and let the FDA reviewer see the reasoning.
Not this
Silently deleting the row to make the report look cleaner.
Cybersecurity is not a brake on MedTech innovation - it is the runway that lets new devices reach patients safely.
We say 'yes, and here is the secure path' more than we say 'no'. We help clients ship AI-enabled imaging, connected wearables, and SaMD by designing security in, not bolting it on at the end.
Scenarios from our work
Situation: An early-stage SaMD startup wants to ship a cloud-connected companion app but has no security program.
What this looks like
Give them a right-sized SPDF that fits a 6-person team - threat model, SBOM, secure coding checklist - so they can build fast without rebuilding later for FDA.
Not this
Handing them an enterprise-scale program they cannot execute and telling them to come back when they are bigger.
Situation: A product manager wants to add a novel wireless protocol that has no established security guidance.
What this looks like
Threat-model the protocol, document residual risk, and give them a path to submit - not a blanket refusal.
Not this
'That is not standard, so we cannot support it.'
Trust is built by what we deliver, not by what we promise. Every artifact, every meeting, every commitment.
We do what we say we will do, when we said we would do it. If we cannot, we tell the client before the deadline, not after.
Scenarios from our work
Situation: A deliverable is going to slip by three days because a lab test surfaced a deeper issue.
What this looks like
Email the client 72 hours before the original due date with the reason, the revised date, and the mitigation - so they can replan.
Not this
Delivering late with an apology and no advance warning.
Situation: A prospect asks us if we have done exactly their kind of device before.
What this looks like
Say honestly what we have and have not done, and how we would de-risk the gap.
Not this
Overstating experience to win the deal.
The cheapest vulnerability to fix is the one caught in design. The most expensive is the one caught in the field.
We push threat modeling, SBOM, and secure coding upstream. We push postmarket monitoring and coordinated vulnerability disclosure so a device stays safe for its full lifecycle - not just at submission.
Scenarios from our work
Situation: A client's device has been on the market for three years and no one has looked at its SBOM since launch.
What this looks like
Recommend an IEC 81001-5-1 lifecycle refresh: regenerate the SBOM, re-run VEX triage, update the CVD process, and file a postmarket update if warranted.
Not this
Only quoting them when the FDA sends a warning letter.
Situation: A client wants to skip threat modeling because 'the device is simple'.
What this looks like
Do a scoped STRIDE session anyway - 90 minutes catches more than the client expects and prevents a costly FDA deficiency letter.
Not this
Agreeing to skip and hoping FDA does not ask.
FDA guidance, AAMI SW96, and IEC 81001-5-1 are the minimum bar - real security work starts above them.
Passing an FDA review is table stakes. We deliver artifacts that would still hold up if the device were attacked in the field, audited by an EU MDR notified body, or reviewed by a hospital security team.
Scenarios from our work
Situation: A client asks for a 'checkbox' threat model that will pass the FDA reviewer but nothing more.
What this looks like
Deliver a threat model that meets FDA, AAMI SW96, and IEC 81001-5-1 - and that the client's engineers can actually use to fix defects.
Not this
A 12-page STRIDE spreadsheet with no data flow diagram and no mitigations traced to design.
Situation: An engagement passes the letter of FDA guidance but leaves an obvious hospital-network exposure unaddressed.
What this looks like
Flag it, recommend the mitigation, and document the residual risk even if it is out of scope for the current submission.
Not this
'It is out of scope, not our problem.'
Situation: A prospect wants a checkbox package to clear the FDA and pushes back when we raise residual risk to patients.
What this looks like
Decline the engagement. We have turned down multiple prospects who framed cybersecurity as a submission hurdle rather than a patient-safety obligation - and we will keep doing it.
Not this
Taking the revenue, softening findings that have clinical impact, and calling it 'meeting the client where they are.'
The best cybersecurity outcomes come from working with the client's engineers, regulatory team, and quality team - not around them.
We embed with the client. We teach as we deliver. We treat the client's team as partners, and we treat our own team as one team.
Scenarios from our work
Situation: A client engineer disagrees with a threat model finding.
What this looks like
Sit down together, walk the DFD, and either update the model or update the engineer's understanding - whichever the evidence supports.
Not this
'The report is final, take it up with your PM.'
Situation: Another BGC teammate is behind on their part of a shared deliverable.
What this looks like
Offer help, cover a section, or flag the risk to the PM early - the client sees one BGC.
Not this
'That's their piece, not mine.'
Patients' lives depend on our work. That standard applies to every report, every meeting, every email.
We do not ship sloppy work. We proofread. We fact-check citations. We rerun the tool one more time before we deliver.
Scenarios from our work
Situation: A report is 'good enough' at 5pm on Friday.
What this looks like
One more pass Monday morning to catch the typo in the executive summary and the stale CVE reference in the appendix.
Not this
Sending it Friday because it is done enough.
Situation: A client asks a question and we do not know the answer.
What this looks like
'I do not know - I will confirm and get back to you by end of day tomorrow,' followed by an actual answer.
Not this
Guessing an answer to look confident.