Applied Agent Cybersecurity & Open Agent Kit (OAK)

Hello everyone!

We’re excited to announce the next event of the Vienna AI Engineering Meetup on Tuesday, June 3rd! As always, we’ll be generously hosted by Greentube at their HQ on Wiedner Hauptstraße.

We’ll kick things off with Daniel Noszian, who will speak about a growingly important topic: Cybersecurity in a World of ubiquitous Agents. Then, Valentin Schreibmaier will introduce us to a new open-source tool made in Austria: The Open Agent Kit (OAK).

Come join us as we explore the future of AI Engineering together!

-------------------------------

Here are the detailed abstracts:

Staying safe in the battle of the standards: MCP, A2A, ACP - Applied Cybersecurity for agents and beyond (Daniel Noszian)

The AI agent revolution is upon us, but behind the scenes a high-stakes battle is unfolding over how these digital entities will communicate. Like choosing between VHS and Betamax (but with potentially world-changing consequences), the protocols we adopt today will shape the AI landscape for years to come.
This session will dissect the three contending protocols reshaping our digital future in 2025: Google's Agent-to-Agent (A2A), Anthropic's Model Context Protocol (MCP), and IBM's Agent Communication Protocol (ACP). We'll explore how each represents fundamentally different visions for AI architecture from MCP's "USB-C port for AI models" approach to A2A's vision of autonomous agent ecosystems.
But this isn't just about technical specifications. It's about security implications that could make or break your organization's AI strategy. We'll examine:
- The unique attack surfaces each protocol introduces
- Practical vulnerability assessment strategies for multi-agent systems
- Real-world security trade-offs between centralized vs. distributed agent architectures
- Implementation safeguards that should be non-negotiable, regardless of protocol choice
- Strategic considerations for building resilient, future-proof agent ecosystems
Whether you're actively implementing agent systems or preparing your security posture for the inevitable AI transformation, this session provides the cybersecurity roadmap you need to navigate the protocol wars safely.

About the speaker:

• Daniel is a cybersecurity strategist who bridges the worlds of emerging technologies and enterprise defense. As Practice Lead for AI Strategy & Risk at SEC Consult, he helps organizations navigate the labyrinth of artificial intelligence opportunities while keeping the monsters of security vulnerabilities at bay.
  With over 25 years in the IT and cybersecurity trenches, Daniel currently leads Security Operations Consulting at SEC Consult, where he architects resilient security ecosystems for international companies and public sector entities. His expertise spans strategy, incident response frameworks, threat modeling, and the art of translating technical jargon into business strategies that executives actually understand.
  Before diving into the security consulting pool, Daniel served as CTO for an online higher education provider, where he transformed traditional learning environments into cloud-based ecosystems while implementing governance frameworks that kept both auditors and students happy.
  Daniel's career journey through various cybersecurity and IT strategy roles has equipped him with the rare ability to see both forest and trees – connecting technical implementations with business objectives while keeping risk management at the core of digital transformation.

The Open Agent Kit - OAK (Valentin Schreibmaier)
In this presentation, we will explore the development and capabilities of the Open Agent Kit (OAK), an open-source platform designed to streamline the creation and deployment of custom AI agents. OAK empowers developers and non developers to build modular, scalable, and production-ready agents that integrate seamlessly with various large language models (LLMs), including OpenAI, Anthropic, and self-hosted alternatives.
We’ll delve into the architecture of OAK, highlighting its support for plugin-based extensibility, built-in knowledge base ingestion (supporting formats like PDFs, CSVs, and JSON), and embeddable chat interfaces.
Additionally we'll talk about the importance of a PaaS solution for enterprise use cases and the problems of implementation fragmentation within organizations.

About the speaker:

• Valentin is the Founder and CTO of arc nine, a company focused on leveraging AI to empower businesses. Prior to this, he founded and led a development studio in California that was successfully acquired. The studio collaborated with major clients such as Google, HP, Lyft, Nike, DirecTV, ESPN, and Pepsi building web applications that served millions of customers.

-------------------------------

Schedule
18:00 - Get together with food and beverages
18:30 - "Staying safe in the battle of the standards: MCP, A2A, ACP - Applied Cybersecurity for agents and beyond" (Daniel Noszian)
19:15 - "The Open Agent Kit - OAK " (Valentin Schreibmaier)
20:00 - More drinks & networking.

Looking forward to seeing you at the Greentube HQ!

-------------------------------

This event is sponsored by Greentube, a NOVOMATIC company.

NOVOMATIC is the leading provider of gaming technology and casino equipment in Europe. Sponsoring the Vienna AI Engineering Meetup is part of our effort to contribute to the creation of a strong Data & AI community in and around Vienna.