On October 22-23, 2026, the San Jose McEnery Convention Center will host AGNTCon + MCPCon North America. With over 3,500 attendees and 150 talks, the event marks a shift in the agent economy: the transition from experimental, isolated scripts to a standardized, institutionalized infrastructure layer. Hosted by the Agentic AI Foundation under the Linux Foundation, this is the first major conference dedicated specifically to agent infrastructure and the Model Context Protocol (MCP).
The thesis here is straightforward: the chaotic proliferation of agent tools is being reined in by a formal governance stack. While early agent development felt like a “wild west” of proprietary integrations, the industry is now coalescing around shared protocols. What matters here is not just the technology, but the emergence of a consensus on how these systems should talk to one another, secure their boundaries, and verify their actions.
The conference agenda serves as a map of this emerging stack. By organizing 11 tracks around themes like security, evals, interoperability, and multi-agent systems, the organizers are effectively codifying the requirements for enterprise-grade agent deployment. The speaker lineup further confirms this structural focus. David Soria Parra, co-creator of MCP at Anthropic, and Rao Surapaneni of Google Cloud—who launched the A2A protocol—represent the foundational communication layers. Meanwhile, Bradley Axen of Block, creator of the Goose framework, addresses the runtime authority layer, and Manik Surtani of the Agentic AI Foundation provides the institutional oversight necessary for open-source adoption.
This event directly mirrors the governance stack Forkast has been tracking. Our previous analysis of the MCP Governance Surface, the MCP Platform Battleground, and the Google Managed Agent Harness all pointed toward this moment of consolidation. The inclusion of workshops on agent gateways and access boundaries suggests that the industry is moving past the “can we build it” phase and into the “how do we govern it” phase.
For infrastructure builders, the most telling signal is the Solo.io keynote, which explicitly addresses the migration of agents from local desktops to shared, scalable infrastructure. This is the critical pivot point for the agent identity layer; without shared infrastructure, agents remain tethered to individual user machines, limiting their utility in complex, multi-party workflows. The introduction of the MCPA (Model Context Protocol Associate) certification at the conference further signals that the market is ready to standardize the skills required to manage this new stack. We know that standardization is the precursor to mass adoption, but we are still in the early stages of determining which protocols will dominate. While the Agentic AI Foundation is positioning itself as the central clearinghouse for MCP and A2A, the actual implementation across diverse enterprise environments remains an unresolved variable. We are watching to see if these standards can remain flexible enough to accommodate rapid innovation while being rigid enough to satisfy security and compliance requirements.
The deeper issue is whether the current governance stack can handle the transition from human-in-the-loop systems to autonomous, multi-agent ecosystems. As speakers like Sarah Drasner discuss the “Agentic Web” and Dawn Song addresses AI safety, the focus is clearly shifting toward the long-term stability of these systems. The conference is not just a gathering of developers; it is a formalization of the rules of engagement for the next generation of digital labor.
Looking ahead, the success of this conference will be measured by the adoption of these protocols outside of the core contributors. If the tracks on interoperability and security lead to tangible, cross-platform implementations, we will have moved from a collection of disparate agent tools to a cohesive, functional agent economy. For now, the infrastructure is being laid; the next step is seeing who actually builds on top of it.