Agent Interoperability Workshop

Join us in person to explore how different agents can work together seamlessly at the Agent Interoperability Workshop!

Agent Interoperability Workshop

AI agents are starting to show up everywhere: in products, developer tools, internal workflows, and the everyday experiments people run on their own machines. We already have decent answers for connecting agents to software through MCP, CLIs, APIs, and tool calling, and for coordinating agent teams inside a single application or organisation through harnesses and orchestration. But what happens when agents built with different designs and by different owners try to find each other and communicate directly across the internet? That is the question this workshop explores.

Hosted by Shannon Gray of Gray Sky AI and Samir Ghanem of Brace Cyber as part of AI Week in Melbourne, Victoria, this hands-on workshop is for AI engineers and enthusiasts who want to move beyond demos. You will see what agent interoperability actually requires: discovery, peer-to-peer authentication, structured messages, shared session context, encryption, and enough trust machinery for independent systems to work together across trust boundaries.

You will work through a practical agent-to-agent interoperability lab using a prototype interface based on the IEEE P3394 draft standard for agent-to-agent interoperability. This is not a lecture-only session: you will run commands, inspect manifests, create local identity material, serve or call an agent, and pair with other attendees to complete a signed agent-to-agent round trip.

What You Will Do

• Start from a safe hosted workspace with the agent environment and interface ready, so you can focus on interoperability rather than infrastructure setup.
• Learn the core building blocks of agent interoperability: manifests, channels, adapters, messages, principals, relationships, and sessions.
• Run your hosted agent or client and verify each step with concrete command output.
• Inspect public discovery metadata, endpoint details, manifests, and pinned keys.
• Set up local identity and encryption material for signed agent-to-agent communication.
• Explore practical security questions around trust, identity, and cross-organisation agent communication.
• Put the pieces together by experimenting with real agent-to-agent interoperability, including 1:1 flows with other attendees.
• Join the shared arena to see what changes when larger numbers of independently owned agents have to coordinate across trust boundaries.
• Leave with a working mental model for how agent systems can communicate beyond a single framework or application boundary.

Who This Is For

This workshop is designed for software developers, AI engineers, technical founders, solution architects, and technically hands-on product or platform people who are already comfortable working in a terminal.

It is especially relevant if you are experimenting with AI agents, building internal automation, evaluating multi-agent architectures, or thinking about how agents should communicate across teams, tools, vendors, and organisations.

Requirements

This is a practical coding workshop. You do not need to be an expert in agent interoperability or networking concepts, but it helps to arrive with enough development fluency to keep up with guided command-line work.

Required:

• Comfortable reading and running terminal commands.
• Ability to self-serve and troubleshoot teething issues.
• Working knowledge of Python, including virtual environments, packages, and running Python modules or scripts.
• Familiarity with AI agents, LLM tool use, and agentic workflows.
• A laptop where you can install or run developer tools. We recommend using a personal, test, or otherwise isolated development machine rather than a locked-down production or corporate device.
• A modern browser and a code editor.

Recommended:

• Familiarity with HTTP APIs, JSON/YAML, and local development servers.
• A backup internet option, such as a phone hotspot, in case venue Wi-Fi is unstable.
• A powerbank or other backup power option as you may not have reliable access to a power outlet.

You will get the most out of the session if you have built or modified Python applications and have at least experimented with AI agents or tool-calling LLMs. If you are brand new to coding or have not used a terminal before, please let the organisers know before attending, we will do our best to accomodate everyone as best we can.

Format

This is a hands-on workshop of up to 6 hours with facilitator support. The main track in the guide takes around 2-4 hours, leaving time for setup help, questions, pairing, session work, and optional stretch activities.

#Melbourne #AIWeek #AgentInteroperability

Event Rules

Please read before purchasing a ticket: event_rules

Important Note

This workshop uses an unofficial P3394 interoperability prototype for education, exploration, and implementation feedback. It is not an official IEEE product, reference implementation, or training session.