Client Outcomes & Insights

What Defensible Security
Actually Looks Like.

Real engagements. Real findings. Real outcomes. Every case study below reflects work delivered by our senior team — anonymized to protect client confidentiality, specific enough to be useful.

3
Avg. Critical Findings Per Engagement
45%+
Avg. Critical Risk Reduction Post-Remediation
100%
Senior-Led — No Junior-Only Delivery
1 Day
Avg. Follow-Up After Scoping Call
K–12 Education Full Threat Risk Assessment

Full TRA Gave a Large School Division the Risk Clarity Their Board Had Been Asking For

A large Canadian school division — responsible for thousands of students, staff, and sensitive personal data — needed a comprehensive Threat Risk Assessment that could support board-level risk governance, inform security investment decisions, and stand up to external scrutiny. Previous assessments had been too surface-level to drive real action.

Full asset-based threat modeling completed — covering infrastructure, third-party systems, student data flows, and administrative access across the division
Prioritized risk register delivered — likelihood and impact scored across all identified threats, with treatment options documented for each risk so leadership had clear choices, not just findings
Board-ready executive summary produced — non-technical narrative translating risk findings into governance language, enabling trustees to make informed decisions on security investment

Our trustees had been asking for a clear picture of our security risks for two years. What 3Tenets delivered wasn't just a risk list — it was a decision-making tool. For the first time, we could walk into a board meeting and explain our risk posture with confidence.

DS
Director of Technology
Large Canadian School Division
Municipal Government Penetration Testing • NIST Assessment • Full TRA

One Integrated Engagement. Three Deliverables. A Complete Security Posture Picture.

A Canadian town needed penetration testing, a NIST CSF 2.0 maturity assessment, and a full Threat Risk Assessment — and needed them to tell a coherent story rather than exist as three disconnected reports. With council scrutiny on IT spending and a procurement approval deadline in sight, there was no room for a generic, checkbox-driven approach.

All three workstreams delivered as a unified program — findings from the pentest directly informed the TRA risk register, and the NIST assessment provided the maturity baseline for remediation sequencing
Active vulnerabilities surfaced through testing — not just theoretical risks, but real exploitable paths in the town's network and public-facing systems, confirmed with evidence
Procurement-ready documentation produced — the combined output satisfied council approval requirements and gave the CAO defensible documentation for the public record

We'd had vendors propose running these as three separate engagements over 18 months. 3Tenets understood that a municipality our size needed this done as one coordinated program — and the integrated output was far more useful than three standalone reports would have been.

CA
Director, Information Security
Canadian Town, Ontario
Healthcare Executive Tabletop Exercise

Executive Tabletop Revealed the Real Crisis Was Decision Authority, Not the Incident Response Plan

A regulated Canadian healthcare institution engaged 3Tenets to facilitate a senior leadership tabletop exercise covering cybersecurity incident response, communications protocols, and disaster recovery. Leadership expected the exercise to validate plans already in place. What it found was more valuable — and more urgent.

Decision authority gaps exposed — no clear owner for critical containment calls during the scenario; multiple executives assumed someone else held authority to act
Documented RTO vs. clinical reality gap confirmed — disaster recovery plan assumed a 4-hour system recovery; exercise revealed actual clinical workflow dependencies required a minimum of 12
External communications plan untested — no pre-approved messaging framework existed for simultaneous patient, regulator, and media notification under breach conditions
14 prioritized recommendations delivered across IR, communications, and disaster recovery — each with assigned owner, effort estimate, and 30/60/90-day implementation timeline

We walked in thinking we had a reasonable plan. We walked out knowing exactly where it would have failed us — and with a clear path to fix it. That's the value of an exercise designed to stress-test, not just validate.

DT
Director of Technology
Regulated Canadian Healthcare Institution
Higher Education & Research AI & LLM Security Research — Mitacs Partnership

74% Prompt Injection Success Rate. Found Before a Healthcare AI Reached Clinical Use.

In partnership with the University of Guelph's School of Computer Science under the Mitacs program, 3Tenets served as industry partner on a research engagement to security-test a locally-hosted Llama 3 medical AI agent built with Ollama. The model was designed to simulate a healthcare chatbot handling sensitive patient queries. Before any clinical consideration, the research team needed rigorous adversarial validation — and a framework the institution could carry forward for future AI deployments.

74% attack success rate confirmed via NVIDIA Garak — using the HijackHateHumans probe, 18 injection attempts were executed against the model; 8 resulted in critical compliance failures, demonstrating the model's susceptibility to malicious prompt override in a healthcare context
4 distinct attack classes validated — Direct Command Override, Casual Instruction Injection, Conversational Diversion, and Emergency Stop Commands all achieved high success rates, bypassing the model's safety behaviours across different prompt styles
Full OWASP LLM Top 10 evaluation framework applied — findings mapped to recognized AI vulnerability classifications, producing research-grade documentation suitable for academic publication and institutional AI governance
Actionable defence framework delivered to the university — recommendations included API gateway instruction filtering, mandatory safety prefixes, output validation layers, and model fine-tuning on injection resistance — giving the institution a repeatable security posture for future AI deployments

The Garak scan results were eye-opening. A 74% attack success rate on a model we intended for healthcare use made clear that AI security testing can't be an afterthought. 3Tenets brought the industry expertise that turned our academic research into something with real-world defensibility — the kind of rigour that matters when patient data is involved.

WZ
Dr. Wenjing Zhang, Assistant Professor
School of Computer Science, University of Guelph
Perspective

Security Decisions That Hold Up

Observations from the field — what separates assessments that drive action from ones that collect dust.

Ready to see what defensible security looks like for your organization?

In a 30-minute scoping call, we'll tell you exactly what we'd assess, how we'd approach it, and what you'd receive — no obligation, no sales pressure.