// ADVISORY — ARTIFICIAL INTELLIGENCE × CYBERSECURITY
Understand what your AI can access, define what it is allowed to do, and test whether its controls work. We bring secure engineering, governance, and cybersecurity together so you can scale with confidence.
Founded by security and AI leaders formerly of Nike, Cisco, and Scale AI, with active AI research at the University of Chicago.
P—01 / 05 Build
Move to explore · click to rotateTap cube to change service
We design and deliver useful AI capability within secure architecture, trusted data, defined authority, and accountable workflows.
We help leaders make sound AI decisions and establish an operating model that enables innovation without unmanaged enterprise risk.
We secure AI systems against adversarial misuse and apply AI to increase cyber team capacity, speed, coverage, and decision quality.
We establish evidence that an AI system or agent is sufficiently secure, controlled, traceable, and reliable for its intended business use.
We equip leaders, builders, defenders, and the wider workforce with practical skills to adopt, govern, secure, and respond to AI responsibly.
Who you'll work with
Previously led Global Cyber Threat Intelligence at Nike and built Cisco’s Cyber Threat Analytics & Incident Response practice. Executive MBA, Harvard.
Postdoctoral Scholar, University of Chicago. Former AI Strategy Consultant, Scale AI. Agent-benchmark builder; published at NeurIPS, ICML, and ICLR.
Why NextPuzzleAI
We're not an audit firm, so auditor-independence rules don't limit the AI advisory we can provide.
We sell no security products and recommend controls based only on your environment.
You work directly with the founders, not junior staff.
A specialist firm focused on securing, governing, and testing AI, not a generalist consultancy with an AI add-on.
Research & insights
Professional-reasoning benchmark for evaluating models on high-stakes finance tasks.
View ↗ PAPER · SCALE AIRubric-based evaluation of deep-research agents.
View ↗ BENCHMARKMeasuring how well AI agents complete real, paid remote work.
View ↗ SECURITY BENCHMARK · APPENSecurity benchmark for evaluating AI systems against adversarial attacks.
View ↗Contact
We're a small, hands-on team, and every note reaches us directly — nothing disappears into a queue. Tell us a little about what you're building or defending. A founder will reply within one business day to arrange your discovery call.
Prefer email? hello@nextpuzzleai.com · Reporting a vulnerability? security@nextpuzzleai.com