The Agentic Shift: Why Standardized Foundations Are the Only Path to Scale

The move to agentic AI is a fundamental change in how organizations operate, not just a technical upgrade. While most teams focus on the excitement of new model releases, the real competitive advantage comes from the unglamorous work of building neutral, open-source standards. Angie Jones, VP of the Agentic AI Foundation, explains that the biggest barrier to widespread adoption is not model capability, but a chaotic lack of interoperability. For leaders, the path is clear: those who invest in neutral infrastructure now will avoid the vendor lock-in that stalls innovation. By prioritizing open standards like the Model Context Protocol (MCP) today, organizations can build flexible systems that remain effective as the ecosystem changes, future-proofing their operations against the rapid pace of the next 18 months.

The Hidden Cost of Move Fast and Break Things

In the early days of agentic AI, teams often rushed to use proprietary solutions, treating every integration as a custom engineering project. This creates an immediate problem: high maintenance costs and fragmented workflows. As systems grow from single agents to thousands, this complexity compounds. Jones notes that the industry has moved from simple, single-agent tasks to complex, multi-agent orchestration in a matter of months.

When organizations build on proprietary, closed-loop systems, they trade long-term agility for short-term speed. The result is that teams become workers for the agents, spending more time debugging fragile, non-interoperable connections than delivering value.

If you wanted to use one of these projects but it lives in one of these companies, you might be a bit hesitant as a company to build your products on top of this. If it lives with this one company, I don't know what they are gonna do. I don't know if they will like kill it.

-- Angie Jones

The 1% Rule: Scaling Knowledge Through Champions

The most common failure in organizational AI adoption is the firehose approach, where companies try to upskill everyone at once. Jones argues that this ignores how organizations actually work. Instead of a top-down mandate, she suggests the 1/9/90 rule: identify the 1% of creators across different departments, give them dedicated time, and task them with embedding agentic workflows directly into the system.

This creates a feedback loop where these champions filter out the noise of the AI hype cycle. By baking agent skills and context engineering directly into the codebase, the system itself becomes the teacher. This shifts the burden of education from constant, manual training to a self-sustaining infrastructure that benefits the entire organization.

Why Competitors Collaborate on the Game Board

The formation of the Agentic AI Foundation shows how the industry is handling foundational technology. While companies like OpenAI, Anthropic, and Block are fierce competitors, they recognize that they must collaborate on the rules of the game.

They actually are very friendly with each other. They might say, yeah, our companies just got into this heated Twitter war or something. But you and I have to figure this standard out. You can let those crazy people talk, but we have to get the thing done right now.

-- Angie Jones

This is a systems-level insight: when the underlying infrastructure, such as MCP or A2A, is neutral, competition shifts from who owns the plumbing to who provides the best value on top of it. This creates a lasting advantage for the industry by preventing the need to reinvent the wheel for every new security, identity, or interoperability requirement.

Key Action Items

• Audit your dependency chain: Over the next quarter, identify which agentic tools are locked into proprietary vendor ecosystems. Move toward open standards like MCP to ensure your infrastructure remains portable.
• Implement the 1/9/90 model: Within the next 30 days, identify the 1% champions in your engineering organization. Dedicate 30% of their time to embedding AI context and agent skills directly into your core repositories rather than relying on external training sessions.
• Shift from AI-first to Human-advantage mapping: Audit your current workflows. Identify tasks where human intuition still outperforms current agents. This pays off in 12 to 18 months by preventing the over-delegation of high-stakes decisions to immature systems.
• Join the neutral table: If your organization is building agentic systems, participate in the open working groups of the Agentic AI Foundation. This is a long-term investment that provides early access to standard-setting, preventing future technical debt.
• Prioritize interoperability over features: When selecting tools, prioritize those that support the A2A (Agent-to-Agent) protocol. This ensures that as your agent count scales from dozens to thousands, they can coordinate rather than operating in silos.