An update from AAIF’s new Governing Board Chair, David Nalley
David Nalley, AAIF Governing Board chair and Director of Developer Experience at Amazon Web Services
In less than three months since the Linux Foundation formed the AAIF, we have seen an avalanche of interest from organizations joining to invest in the advancement of agentic AI infrastructure. The news today that 97 companies signed up as new Gold and Silver members at the first opportunity to join, is astounding. That’s more than double the number of companies that joined the Cloud Native Computing Foundation (CNCF) in its first six months – a wildly successful ecosystem that now boasts nearly 800 CNCF members. This interest reflects both the explosive growth of the emerging AI space and the robust network and level of trust the Linux Foundation has built over two decades.
First, thank you to all of the new AAIF members for your votes of confidence in the foundation. Contributing your limited resources to advance common infrastructure is a declaration that open source collaboration works. Our members aren’t just making financial commitments—they’re contributing code, participating in working groups, and advancing agentic protocols and tools. This collaborative approach will accelerate agentic AI for everyone. By combining knowledge from diverse organizations, we can solve complex technical challenges and build tools that serve developers, infrastructure operators, and end users.
I’m personally thrilled to see so many companies using agentic AI to automate and enhance their core business processes, join this effort. Among the new Gold members we have financial services firms like American Express, Circle, JPMorgan Chase, and Worldpay applying agents to fraud detection, trading, and risk assessment; automation and software companies like Autodesk, Diagrid, Keycard, UiPath and Workato that are making their platforms more intelligent and autonomous; and hardware and digital infrastructure providers like Akamai, Equinix, Hitachi, Huawei, Infobip, Lenovo, Red Hat, ServiceNow, and TELUS that are using agents for network optimization, edge computing, and service management.
Technical challenges ahead
These companies are joining us at a critical time for the AAIF and for agentic technologies. Agents are rapidly moving from experiments on developers’ laptops to becoming production-ready at scale. We are, essentially, designing a new distributed computing system, built on the solid foundation of open source cloud native technologies.
Many challenges remain, however, that will require industry-wide collaboration. For example, complex orchestration is required to enable enterprise users to move beyond simple training workloads to build fleets of agents that work together like microservices. This requires sophisticated resource allocation for specialized hardware (GPUs) that foundational cloud native technologies such as Kubernetes weren’t originally designed for but have since adapted to support. Our infrastructure needs to evolve with new AI-centric workflows and modes of computing.
Similarly, agentic technologies must adapt to production needs. MCP’s dynamic client-server approach creates new security challenges, for example, as malicious actors can trick users into granting unauthorized access, as Paul Carleton, MCP core maintainer and a technical staff member at Anthropic, explained at MCP Summit Europe last October. Unlike traditional OAuth workflows where developers pre-register applications with services through manual processes, MCP requires dynamic connections between unknown clients and servers. This eliminates the high-trust relationship established through upfront registration, creating authorization issues that MCP maintainers are actively addressing through proposed solutions such as OAuth client ID metadata documents (SEP-991), registry-based trust, and conformance testing and tooling.
As a foundation now 146 members strong, the AAIF will play an important role in addressing these challenges and making sure that agents can interoperate with each other, and with common tools and frameworks at scale. We must also work to ensure that the distributed infrastructure that was built and optimized for containerized application development can rapidly adapt to help make agents more secure, easier to troubleshoot and debug, and more reliable for large workloads. This will be a massive, industry-wide undertaking that the Linux Foundation, and the AAIF under its purview, is uniquely positioned to orchestrate as a vendor-neutral mediator that spans multiple technologies and ecosystems.
Early Wins: MCP Apps Success
We’ve already seen significant progress under the AAIF with the launch of MCP Apps on Jan. 26, the first official Model Context Protocol (MCP) extension that brings interactive UI capabilities to MCP clients. Most notable was the partnership between OpenAI and Anthropic in creating shared open standards—a collaboration made possible by MCP’s move to the neutral governance structure of the AAIF. Building on OpenAI’s Apps SDK and the community-driven MCP-UI project, the two companies worked together to develop MCP Apps as a unified standard that enables tools to return rich, interactive interfaces—dashboards, forms, visualizations, and multi-step workflows—that render directly in conversations.
The collaboration has yielded impressive adoption, with ChatGPT, Claude, goose, and Visual Studio Code all shipping support for the feature, allowing developers to create interactive experiences that work across multiple widely adopted clients without writing client-specific code. As Nick Cooper from OpenAI noted, “We’re proud to support this new open standard,” while David Soria Parra from Anthropic expressed excitement about “what the community will build.”
Governance milestones
As foundation leaders, our job is to ensure that critical AI technologies such as MCP, goose, and Agents.md evolve transparently and collaboratively rather than under the proprietary control of a few companies. The first step in this process has been establishing a functioning Governing Board to begin setting the foundation’s strategic direction and the governance structure for the projects under its care. The formal governance models for MCP, goose and Agents.md will define the projects’ structure, processes for implementing changes, and contribution guidelines.
Since the AAIF launch on Dec. 9, industry leaders from the eight founding platinum organizations have met regularly along with Gold and Silver tier participants and Linux Foundation leadership. We discuss topics such as the roles of the Technical Committee vs. the Governing Board in decision-making, which sub-committees are most needed, and the timing and criteria for accepting new technical projects.
The board has appointed a Technical Committee (TC) with one representative from each of the eight Platinum members to define the requirements for new projects joining the Foundation and to review new project proposals. The Technical Committee holds biweekly meetings that are recorded and available for anyone to attend and observe on the LFX platform. The Governing Board also established seven new working groups focused on “identity and trust”, “accuracy and reliability”, “workflows and process integration”, “agentic commerce”, “security and privacy”, “observability and traceability”, and “governance, risk and regulatory alignment”. Attendance at working group meetings has been strong in their first week.
The working groups are each chaired by a platinum member and are tasked with generating proposals for the Governing Board and TC on how to solve problems in their areas of focus. For example, the identity and trust working group will address how agentic systems manage identities, establish trust for transactions, and enable agent autonomy while maintaining a human in the loop. The working groups are open to all levels of membership for participation and are chaired by members from our Platinum and Gold tiers.
In January, the board also welcomed Zhong Wu, VP of applied ML engineering at Shopify, to represent Gold members in the Governing Board, further strengthening the foundation’s commitment to inclusive, community-driven governance. And today I’m pleased and humbled to accept the new role of board chair for the AAIF Governing Board.
In coming weeks and months, I expect the Governing Board to further stabilize the new foundation and governance structure, in preparation to welcome new projects to the foundation. Given the incredible interest and enthusiasm we have received from members excited to participate and bring their technologies to the AAIF, we expect to see more exciting announcements on the new AAIF blog going forward. As the Governing Board decides which areas to focus on, we also want to hear from our members and the broader community about the challenges and opportunities you see for agents going forward.
Join the conversation on our Discord community and follow us on X, LinkedIn, and Bluesky.
Finally, please join us on Oct. 22-23 in San Jose for the first AGNTCon + MCPCon, the AAIF’s new flagship annual conference for the open agentic AI ecosystem. Registration opens in March. Also don’t miss MCP Dev Summit North America in New York, NY from April 2-3 and MCP Dev Summit Europe in Amsterdam, The Netherlands from September 17-18.