You shipped OT ZTNA in March. You are hiring AI/ML researchers with PhDs and publication records. Risk Sentinel is on the homepage as an intelligent scoring engine. This page is the result of reading all of that and testing one hypothesis about the gap between your AI research posture and what a CISO will actually buy in 2026.
Appgate leadership came back after the recapitalization. The research org is being built out. The GDIT partnership on the Air Force Next Generation Gateway is public. The direct-routed architecture is the differentiator you tell analysts about. The open question sits one layer above the product: how the AI research org turns into shipped, explainable capability the CISO buyer will trust.
Most security vendors treat AI as a feature to announce. The useful question is which problems inside a ZTNA stack belong on which layer. Most of what gets shipped as "AI-powered" is mis-routed across these three.
Rule-based. Deterministic. Works today. Putting an LLM here adds latency without accuracy.
Traditional code and data infrastructure. Works today. Not an AI problem.
Some of this is a genuine AI problem. Some of it is a statistical model that would do the job faster with better guarantees. Some of it should not ship.
The KPMG UK engagement delivered to regulated-industry leadership. KPMG already sells governance for a living. They hired Eduba because the outside methodology let them keep their own voice while moving faster. The format maps to an Appgate audience: leadership, research leads, marketing leadership, and the CISO-facing sales leaders who carry the analyst briefings.
Delivered through Correlation One into Pacific Life and Colgate-Palmolive. Regulated industries, senior and technical audiences, measured outcomes.
Appgate is hiring PhDs who publish. Eduba ships research to match. ICM is the lead when the conversation is agent architecture. Ethics Engine is the backup when it turns to AI governance for regulated customers.
Submitted to ACM TiiS. Agent context structured as a layered filesystem, L0 identity through L4 working artifacts, with measurable interpretability and reproducibility gains. 52-member practitioner community.
A psychometric assessment tool for evaluating ideological and moral patterns in LLMs. Useful when the conversation turns to AI governance for regulated customers, which Appgate sells into by default.
Eight years in the Marine Corps on cryptographic systems and F-35 / F-18 avionics. MSc Future Governance, University of Edinburgh. Published in ACM TiiS (ICM) and arXiv (Ethics Engine). 1,500+ people trained across enterprise engagements since May 2025.
The Marine background matters here because the GDIT and Air Force work has the same posture. Eduba partners with NLP Logix for work that sits below the orchestration layer. NLP Logix has been in machine learning since 2011 and runs over 150 data scientists.
We will map it on a napkin. Rules layer. Code layer. AI layer. Delete list. If the hypothesis does not hold, you get a map of your own stack out of the meeting anyway.
Matt Creamer, Eduba's CRO. On the floor at eMerge Americas all day.